From cd26c0d99cc46a578252e78f5aa76f50df2909fa Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 10:20:49 +0200 Subject: [PATCH 01/51] what it should look like --- src/transformers/utils/generic.py | 42 +++++++++++++++++++++++++++++++ 1 file changed, 42 insertions(+) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index cfeaaa4fd4c6..52cb8f2e6ab5 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -926,6 +926,8 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): + + is_requested_to_return_tuple = kwargs.pop("return_dict", True) is False is_configured_to_return_tuple = self.config.use_return_dict is False if hasattr(self, "config") else False @@ -952,6 +954,46 @@ def wrapper(self, *args, **kwargs): return wrapper +def check_model_inputs(func): + """ + Decorator to check if the model inputs are valid before calling the function. + It raises a ValueError if the inputs are not valid. + """ + + @wraps(func) + def wrapper(self, *args, **kwargs): + kwargs["output_attention"] = kwargs.get("output_attention",self.config.output_attentions) + kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) + kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) + + # Extract inputs + input_ids = kwargs.get("input_ids", None) + inputs_embeds = kwargs.get("inputs_embeds", None) + use_cache = kwargs["use_cache"] + past_key_values = kwargs.get("past_key_values", None) + + # Input validation + if (input_ids is None) == (inputs_embeds is None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds.") + + # Gradient checkpointing warning + if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: + logger.warning( + "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." + ) + kwargs["use_cache"] = False # update it directly in kwargs + + # Past key values type check + if past_key_values is not None and not isinstance(past_key_values, Cache): + raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") + + return_dict = kwargs.get("return_dict", self.config.use_return_dic) + output = func(self, *args, **kwargs) + if return_dict: + output = output.to_tuple() + return output + + return wrapper class GeneralInterface(MutableMapping): """ From 59a90ee20289251ef71be1e14bd337694928f722 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 10:26:38 +0200 Subject: [PATCH 02/51] waht it should look like --- .../models/llama/modeling_llama.py | 58 ++----------------- 1 file changed, 6 insertions(+), 52 deletions(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index 314b5c323b46..68e238d8c048 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -41,7 +41,7 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...pytorch_utils import ALL_LAYERNORM_LAYERS -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging, check_model_inputs from .configuration_llama import LlamaConfig @@ -248,7 +248,7 @@ def forward( if self.config._attn_implementation != "eager": attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] - attn_output, attn_weights = attention_interface( + attn_output = attention_interface( self, query_states, key_states, @@ -261,7 +261,7 @@ def forward( attn_output = attn_output.reshape(*input_shape, -1).contiguous() attn_output = self.o_proj(attn_output) - return attn_output, attn_weights + return attn_output class LlamaDecoderLayer(GradientCheckpointingLayer): @@ -370,7 +370,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -379,31 +379,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -428,25 +407,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -454,26 +422,12 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( - last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + last_hidden_state=hidden_states, past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class LlamaForCausalLM(LlamaPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] From a6cfc4b6c56525998dca510123014a5e8f13e722 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 10:39:01 +0200 Subject: [PATCH 03/51] updates --- .../models/llama/modeling_llama.py | 11 ++---- src/transformers/utils/generic.py | 38 +++++++++++++++++-- 2 files changed, 38 insertions(+), 11 deletions(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index 68e238d8c048..ca7d496b2d05 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -248,7 +248,7 @@ def forward( if self.config._attn_implementation != "eager": attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] - attn_output = attention_interface( + attn_output, attn_weight = attention_interface( self, query_states, key_states, @@ -261,7 +261,7 @@ def forward( attn_output = attn_output.reshape(*input_shape, -1).contiguous() attn_output = self.o_proj(attn_output) - return attn_output + return attn_output, attn_weight class LlamaDecoderLayer(GradientCheckpointingLayer): @@ -309,12 +309,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 52cb8f2e6ab5..ba9f301cc4ef 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -31,6 +31,7 @@ import numpy as np from packaging import version +from ..modeling_layers import GradientCheckpointingLayer from .import_utils import ( get_torch_version, is_flax_available, @@ -966,6 +967,27 @@ def wrapper(self, *args, **kwargs): kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) + hooks = [] + collected_attentions = [] + collected_hidden_states = [] + output_attentions = kwargs["output_attention"] + output_hidden_states = kwargs["output_hidden_states"] + if output_attentions or output_hidden_states: + # GradientCheckpointingLayer is the one we extrct the hidden states from + # it is also the one that returns hidden_states, self_attn_weights + def hook_fn(module, inp, out): + if output_hidden_states: + collected_hidden_states.append(out[0]) + if output_attentions and len(out) > 1: + collected_attentions.append(out[1]) + + for layer in self.named_modules(): + if isinstance(layer, GradientCheckpointingLayer): + # Register the hook only for layers that are instances of GradientCheckpointingLayer + # This is to avoid registering hooks on all layers, which can be expensive + # and unnecessary. + hooks.append(layer.register_forward_hook(hook_fn)) + # Extract inputs input_ids = kwargs.get("input_ids", None) inputs_embeds = kwargs.get("inputs_embeds", None) @@ -988,10 +1010,20 @@ def wrapper(self, *args, **kwargs): raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") return_dict = kwargs.get("return_dict", self.config.use_return_dic) - output = func(self, *args, **kwargs) + outputs = func(self, *args, **kwargs) + + for h in hooks: + h.remove() + + if output_hidden_states: + collected_hidden_states.append(outputs.last_hidden_state) + outputs.hidden_states = tuple(collected_hidden_states) + if output_attentions: + outputs.attentions = tuple(collected_attentions) + if return_dict: - output = output.to_tuple() - return output + outputs = output.to_tuple() + return outputs return wrapper From d70e1c1aa5bd9710bb7f091429d62944cef19d95 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:07:58 +0200 Subject: [PATCH 04/51] fixup --- src/transformers/models/llama/modeling_llama.py | 7 ++++--- src/transformers/utils/generic.py | 17 +++++++++-------- 2 files changed, 13 insertions(+), 11 deletions(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index ca7d496b2d05..7c1564d54643 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -41,7 +41,7 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...pytorch_utils import ALL_LAYERNORM_LAYERS -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging, check_model_inputs +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_llama import LlamaConfig @@ -419,7 +419,8 @@ def forward( hidden_states = self.norm(hidden_states) return BaseModelOutputWithPast( - last_hidden_state=hidden_states, past_key_values=past_key_values, + last_hidden_state=hidden_states, + past_key_values=past_key_values, ) @@ -471,7 +472,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index ba9f301cc4ef..da844d521648 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -32,6 +32,7 @@ from packaging import version from ..modeling_layers import GradientCheckpointingLayer +from ..utils import logging from .import_utils import ( get_torch_version, is_flax_available, @@ -42,6 +43,8 @@ ) +logger = logging.get_logger(__name__) + if is_torch_available(): # required for @can_return_tuple decorator to work with torchdynamo import torch # noqa: F401 @@ -927,8 +930,6 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): - - is_requested_to_return_tuple = kwargs.pop("return_dict", True) is False is_configured_to_return_tuple = self.config.use_return_dict is False if hasattr(self, "config") else False @@ -955,6 +956,7 @@ def wrapper(self, *args, **kwargs): return wrapper + def check_model_inputs(func): """ Decorator to check if the model inputs are valid before calling the function. @@ -963,7 +965,7 @@ def check_model_inputs(func): @wraps(func) def wrapper(self, *args, **kwargs): - kwargs["output_attention"] = kwargs.get("output_attention",self.config.output_attentions) + kwargs["output_attention"] = kwargs.get("output_attention", self.config.output_attentions) kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) @@ -1000,13 +1002,11 @@ def hook_fn(module, inp, out): # Gradient checkpointing warning if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: - logger.warning( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) + logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") kwargs["use_cache"] = False # update it directly in kwargs # Past key values type check - if past_key_values is not None and not isinstance(past_key_values, Cache): + if past_key_values is not None and not isinstance(past_key_values, "Cache"): raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") return_dict = kwargs.get("return_dict", self.config.use_return_dic) @@ -1022,11 +1022,12 @@ def hook_fn(module, inp, out): outputs.attentions = tuple(collected_attentions) if return_dict: - outputs = output.to_tuple() + outputs = outputs.to_tuple() return outputs return wrapper + class GeneralInterface(MutableMapping): """ Dict-like object keeping track of a class-wide mapping, as well as a local one. Allows to have library-wide From 388ee4b9232d30698fb4c3e927ffee3d38be650a Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:16:32 +0200 Subject: [PATCH 05/51] check this out --- src/transformers/models/aria/modeling_aria.py | 62 ++---------------- .../models/bitnet/modeling_bitnet.py | 65 ++----------------- .../models/cohere/modeling_cohere.py | 53 ++------------- .../models/cohere2/modeling_cohere2.py | 4 +- src/transformers/models/csm/modeling_csm.py | 58 ++--------------- .../deepseek_v3/modeling_deepseek_v3.py | 65 ++----------------- .../models/diffllama/modeling_diffllama.py | 62 ++---------------- src/transformers/models/emu3/modeling_emu3.py | 55 ++-------------- .../models/gemma/modeling_gemma.py | 16 ++--- .../models/gemma2/modeling_gemma2.py | 4 +- .../models/gemma3/modeling_gemma3.py | 11 +++- src/transformers/models/glm/modeling_glm.py | 62 ++---------------- src/transformers/models/glm4/modeling_glm4.py | 50 ++------------ .../models/gpt_neox/modeling_gpt_neox.py | 4 +- .../models/granite/modeling_granite.py | 4 +- .../models/helium/modeling_helium.py | 62 ++---------------- .../models/llama/modeling_llama.py | 4 +- .../models/mistral/modeling_mistral.py | 14 +--- .../models/moonshine/modeling_moonshine.py | 11 +--- src/transformers/models/olmo/modeling_olmo.py | 65 ++----------------- .../models/olmo2/modeling_olmo2.py | 58 ++--------------- src/transformers/models/phi/modeling_phi.py | 9 +-- src/transformers/models/phi3/modeling_phi3.py | 7 +- .../models/qwen2/modeling_qwen2.py | 14 +--- .../models/qwen3/modeling_qwen3.py | 9 +-- .../models/starcoder2/modeling_starcoder2.py | 14 +--- src/transformers/utils/generic.py | 6 +- 27 files changed, 109 insertions(+), 739 deletions(-) diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index f501a17eb46a..55bd10d06116 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -32,7 +32,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs from ...utils.import_utils import is_torch_available from ..auto import AutoModel from .configuration_aria import AriaConfig, AriaTextConfig @@ -43,9 +43,6 @@ from torch import nn -logger = logging.get_logger(__name__) - - @use_kernel_forward_from_hub("RMSNorm") class AriaTextRMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): @@ -624,12 +621,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -750,7 +742,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -759,31 +751,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -808,25 +779,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -834,20 +794,10 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) @@ -901,7 +851,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/bitnet/modeling_bitnet.py b/src/transformers/models/bitnet/modeling_bitnet.py index f526802bfca9..af60de1d8133 100644 --- a/src/transformers/models/bitnet/modeling_bitnet.py +++ b/src/transformers/models/bitnet/modeling_bitnet.py @@ -34,13 +34,10 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs from .configuration_bitnet import BitNetConfig -logger = logging.get_logger(__name__) - - @use_kernel_forward_from_hub("RMSNorm") class BitNetRMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): @@ -269,12 +266,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights class BitNetRotaryEmbedding(nn.Module): @@ -364,7 +356,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -373,31 +365,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -422,25 +393,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -448,26 +408,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class BitNetForCausalLM(BitNetPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -516,7 +463,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/cohere/modeling_cohere.py b/src/transformers/models/cohere/modeling_cohere.py index 88ca4e31de10..148a2e8e640d 100644 --- a/src/transformers/models/cohere/modeling_cohere.py +++ b/src/transformers/models/cohere/modeling_cohere.py @@ -42,13 +42,10 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs from .configuration_cohere import CohereConfig -logger = logging.get_logger(__name__) - - class CohereLayerNorm(nn.Module): def __init__(self, hidden_size=None, eps=1e-5, bias=False): """The hidden size can be a tuple or an int. The tuple is used for QKNorm to normalize across head_dim""" @@ -401,7 +398,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -410,31 +407,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -459,25 +435,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -485,20 +450,10 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) diff --git a/src/transformers/models/cohere2/modeling_cohere2.py b/src/transformers/models/cohere2/modeling_cohere2.py index 6999f1632f95..8c675573754b 100644 --- a/src/transformers/models/cohere2/modeling_cohere2.py +++ b/src/transformers/models/cohere2/modeling_cohere2.py @@ -34,7 +34,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging from ...utils.deprecation import deprecate_kwarg from .configuration_cohere2 import Cohere2Config @@ -380,7 +380,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/csm/modeling_csm.py b/src/transformers/models/csm/modeling_csm.py index ad374c86e7a5..244a8bb36bea 100644 --- a/src/transformers/models/csm/modeling_csm.py +++ b/src/transformers/models/csm/modeling_csm.py @@ -36,7 +36,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, ModelOutput, auto_docstring, can_return_tuple, logging +from ...utils import ModelOutput, auto_docstring, can_return_tuple, logging from ..auto import AutoModel from .configuration_csm import CsmConfig, CsmDepthDecoderConfig from .generation_csm import CsmGenerationMixin @@ -401,12 +401,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -579,9 +574,6 @@ def forward(self, hidden_states, cache_position=None): return hidden_states -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring( custom_intro=""" The CsmDepthDecoder Model transformer, with a [`CsmCodebooksHead`] on top, @@ -762,10 +754,8 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" @@ -780,25 +770,6 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -823,25 +794,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -849,20 +809,10 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) diff --git a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py index 9a768bfd9f94..0884c2ab0530 100644 --- a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py +++ b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py @@ -22,13 +22,10 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs from .configuration_deepseek_v3 import DeepseekV3Config -logger = logging.get_logger(__name__) - - @use_kernel_forward_from_hub("RMSNorm") class DeepseekV3RMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): @@ -489,12 +486,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -554,7 +546,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -563,31 +555,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -612,25 +583,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -638,26 +598,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class DeepseekV3ForCausalLM(DeepseekV3PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -706,7 +653,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 43e8e4584bdc..8040534544cb 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -48,7 +48,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_diffllama import DiffLlamaConfig @@ -536,12 +536,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -636,7 +631,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -645,31 +640,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -694,25 +668,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -720,26 +683,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class DiffLlamaForCausalLM(DiffLlamaPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -788,7 +738,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index 105bcb4f22be..ef078dadfbd8 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -39,7 +39,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_emu3 import Emu3Config, Emu3TextConfig, Emu3VQVAEConfig @@ -1218,7 +1218,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -1227,31 +1227,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -1276,25 +1255,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -1302,26 +1270,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class Emu3ForCausalLM(Emu3PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -1371,7 +1326,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py index 1f8da9ed0ece..791e52ba2538 100644 --- a/src/transformers/models/gemma/modeling_gemma.py +++ b/src/transformers/models/gemma/modeling_gemma.py @@ -39,7 +39,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_gemma import GemmaConfig @@ -303,12 +303,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -364,7 +359,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -468,9 +463,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class GemmaForCausalLM(GemmaPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -519,7 +511,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/gemma2/modeling_gemma2.py b/src/transformers/models/gemma2/modeling_gemma2.py index 28a57ab9090b..c21171f5ef3d 100644 --- a/src/transformers/models/gemma2/modeling_gemma2.py +++ b/src/transformers/models/gemma2/modeling_gemma2.py @@ -39,7 +39,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from ...utils.deprecation import deprecate_kwarg from .configuration_gemma2 import Gemma2Config @@ -385,7 +385,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/gemma3/modeling_gemma3.py b/src/transformers/models/gemma3/modeling_gemma3.py index 8d78ca92b4ca..403ed9077ee8 100644 --- a/src/transformers/models/gemma3/modeling_gemma3.py +++ b/src/transformers/models/gemma3/modeling_gemma3.py @@ -38,7 +38,14 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import ModelOutput, auto_docstring, can_return_tuple, is_torchdynamo_compiling, logging +from ...utils import ( + ModelOutput, + auto_docstring, + can_return_tuple, + check_model_inputs, + is_torchdynamo_compiling, + logging, +) from ...utils.deprecation import deprecate_kwarg from ..auto import AutoModel from .configuration_gemma3 import Gemma3Config, Gemma3TextConfig @@ -503,7 +510,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index 2ee6273c00d4..113e0e64b9eb 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -40,7 +40,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_glm import GlmConfig @@ -320,12 +320,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -381,7 +376,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -390,31 +385,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -439,25 +413,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -465,26 +428,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class GlmForCausalLM(GlmPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -533,7 +483,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index 75487c5fccff..58c2c9944e35 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -40,7 +40,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_glm4 import Glm4Config @@ -389,7 +389,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -398,31 +398,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -447,25 +426,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -473,20 +441,10 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) diff --git a/src/transformers/models/gpt_neox/modeling_gpt_neox.py b/src/transformers/models/gpt_neox/modeling_gpt_neox.py index 6b08c27a3061..32533cf5cd60 100755 --- a/src/transformers/models/gpt_neox/modeling_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modeling_gpt_neox.py @@ -24,7 +24,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_gpt_neox import GPTNeoXConfig @@ -337,7 +337,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_in = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/granite/modeling_granite.py b/src/transformers/models/granite/modeling_granite.py index d1d69f9579c0..676f21b5f723 100644 --- a/src/transformers/models/granite/modeling_granite.py +++ b/src/transformers/models/granite/modeling_granite.py @@ -35,7 +35,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_granite import GraniteConfig @@ -386,7 +386,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index 31d9f963049f..6fcce60c5908 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -40,7 +40,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_helium import HeliumConfig @@ -305,12 +305,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -366,7 +361,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -375,31 +370,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -424,25 +398,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -450,26 +413,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class HeliumForCausalLM(HeliumPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -518,7 +468,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index 7c1564d54643..c524b798ffb9 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -248,7 +248,7 @@ def forward( if self.config._attn_implementation != "eager": attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] - attn_output, attn_weight = attention_interface( + attn_output, attn_weights = attention_interface( self, query_states, key_states, @@ -261,7 +261,7 @@ def forward( attn_output = attn_output.reshape(*input_shape, -1).contiguous() attn_output = self.o_proj(attn_output) - return attn_output, attn_weight + return attn_output, attn_weights class LlamaDecoderLayer(GradientCheckpointingLayer): diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py index 10b0019aeeac..f3c1a2837546 100644 --- a/src/transformers/models/mistral/modeling_mistral.py +++ b/src/transformers/models/mistral/modeling_mistral.py @@ -26,7 +26,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, logging from .configuration_mistral import MistralConfig @@ -247,12 +247,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -444,9 +439,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class MistralForCausalLM(MistralPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -495,7 +487,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/moonshine/modeling_moonshine.py b/src/transformers/models/moonshine/modeling_moonshine.py index d8f61fbf5021..88323c4bb3d5 100644 --- a/src/transformers/models/moonshine/modeling_moonshine.py +++ b/src/transformers/models/moonshine/modeling_moonshine.py @@ -41,7 +41,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_moonshine import MoonshineConfig @@ -379,12 +379,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights class MoonshineDecoderLayer(GradientCheckpointingLayer): @@ -678,7 +673,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py index c35988e2b8d2..9786ddeaff07 100644 --- a/src/transformers/models/olmo/modeling_olmo.py +++ b/src/transformers/models/olmo/modeling_olmo.py @@ -20,13 +20,10 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs from .configuration_olmo import OlmoConfig -logger = logging.get_logger(__name__) - - class OlmoLayerNorm(nn.Module): """LayerNorm but with no learnable weight or bias.""" @@ -253,12 +250,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights class OlmoRotaryEmbedding(nn.Module): @@ -345,7 +337,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -354,31 +346,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -403,25 +374,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -429,26 +389,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class OlmoForCausalLM(OlmoPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -497,7 +444,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/olmo2/modeling_olmo2.py b/src/transformers/models/olmo2/modeling_olmo2.py index 8e69f43d3ebc..958a864a37b9 100644 --- a/src/transformers/models/olmo2/modeling_olmo2.py +++ b/src/transformers/models/olmo2/modeling_olmo2.py @@ -20,13 +20,10 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs from .configuration_olmo2 import Olmo2Config -logger = logging.get_logger(__name__) - - @use_kernel_forward_from_hub("RMSNorm") class Olmo2RMSNorm(nn.Module): def __init__(self, hidden_size, eps=1e-6): @@ -351,7 +348,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -360,31 +357,10 @@ def forward( position_ids: Optional[torch.LongTensor] = None, past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, + use_cache: Optional[bool] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - - # TODO (joao): remove this exception in v4.56 -- it exists for users that try to pass a legacy cache - if not isinstance(past_key_values, (type(None), Cache)): - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -409,25 +385,14 @@ def forward( ) hidden_states = inputs_embeds - - # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers[: self.config.num_hidden_layers]: - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **flash_attn_kwargs, @@ -435,26 +400,13 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - hidden_states = self.norm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - return BaseModelOutputWithPast( last_hidden_state=hidden_states, - past_key_values=past_key_values if use_cache else None, - hidden_states=all_hidden_states, - attentions=all_self_attns, + past_key_values=past_key_values, ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class Olmo2ForCausalLM(Olmo2PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -503,7 +455,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index 5edceb27c0da..8b9b15d6a613 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -24,7 +24,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging from .configuration_phi import PhiConfig @@ -343,7 +343,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -454,9 +454,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class PhiForCausalLM(PhiPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -505,7 +502,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index 79703927021f..6a94e0fb3d1f 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -41,7 +41,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, logging from .configuration_phi3 import Phi3Config @@ -499,9 +499,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class Phi3ForCausalLM(Phi3PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -550,7 +547,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py index aaebc3c82bd1..d1dd12916a13 100644 --- a/src/transformers/models/qwen2/modeling_qwen2.py +++ b/src/transformers/models/qwen2/modeling_qwen2.py @@ -26,7 +26,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, logging from .configuration_qwen2 import Qwen2Config @@ -251,12 +251,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -458,9 +453,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class Qwen2ForCausalLM(Qwen2PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -509,7 +501,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 6da044857043..47d045c216a2 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -277,12 +277,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights @auto_docstring @@ -535,7 +530,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py index b0179a518bbf..539db55ffbed 100644 --- a/src/transformers/models/starcoder2/modeling_starcoder2.py +++ b/src/transformers/models/starcoder2/modeling_starcoder2.py @@ -44,7 +44,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, logging from .configuration_starcoder2 import Starcoder2Config @@ -250,12 +250,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states, self_attn_weights class Starcoder2RotaryEmbedding(nn.Module): @@ -448,9 +443,6 @@ def forward( ) -class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... - - @auto_docstring class Starcoder2ForCausalLM(Starcoder2PreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] @@ -499,7 +491,7 @@ def forward( output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, - **kwargs: Unpack[KwargsForCausalLM], + **kwargs: Unpack[FlashAttentionKwargs], ) -> CausalLMOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index da844d521648..b6d950217c86 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -1011,9 +1011,9 @@ def hook_fn(module, inp, out): return_dict = kwargs.get("return_dict", self.config.use_return_dic) outputs = func(self, *args, **kwargs) - - for h in hooks: - h.remove() + if output_attentions or output_hidden_states: + for h in hooks: + h.remove() if output_hidden_states: collected_hidden_states.append(outputs.last_hidden_state) From fbad09c9c080c3c20e1feb26bcff8854dcdc6e3e Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:20:25 +0200 Subject: [PATCH 06/51] push --- src/transformers/utils/generic.py | 32 ++++++++++++------------------- 1 file changed, 12 insertions(+), 20 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index b6d950217c86..0a564b142a8d 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -969,15 +969,19 @@ def wrapper(self, *args, **kwargs): kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) + output_attentions = kwargs["output_attention"] + output_hidden_states = kwargs["output_hidden_states"] + input_ids = kwargs.get("input_ids", None) + inputs_embeds = kwargs.get("inputs_embeds", None) + use_cache = kwargs["use_cache"] + past_key_values = kwargs.get("past_key_values", None) + return_dict = kwargs.get("return_dict", self.config.use_return_dict) + hooks = [] collected_attentions = [] collected_hidden_states = [] - output_attentions = kwargs["output_attention"] - output_hidden_states = kwargs["output_hidden_states"] if output_attentions or output_hidden_states: - # GradientCheckpointingLayer is the one we extrct the hidden states from - # it is also the one that returns hidden_states, self_attn_weights - def hook_fn(module, inp, out): + def output_hidden_and_attention(module, inp, out): if output_hidden_states: collected_hidden_states.append(out[0]) if output_attentions and len(out) > 1: @@ -985,31 +989,20 @@ def hook_fn(module, inp, out): for layer in self.named_modules(): if isinstance(layer, GradientCheckpointingLayer): - # Register the hook only for layers that are instances of GradientCheckpointingLayer - # This is to avoid registering hooks on all layers, which can be expensive - # and unnecessary. - hooks.append(layer.register_forward_hook(hook_fn)) + hooks.append(layer.register_forward_hook(output_hidden_and_attention)) - # Extract inputs - input_ids = kwargs.get("input_ids", None) - inputs_embeds = kwargs.get("inputs_embeds", None) - use_cache = kwargs["use_cache"] - past_key_values = kwargs.get("past_key_values", None) - # Input validation - if (input_ids is None) == (inputs_embeds is None): + if input_ids ^ inputs_embeds: raise ValueError("You must specify exactly one of input_ids or inputs_embeds.") - # Gradient checkpointing warning if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") kwargs["use_cache"] = False # update it directly in kwargs - # Past key values type check if past_key_values is not None and not isinstance(past_key_values, "Cache"): raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - return_dict = kwargs.get("return_dict", self.config.use_return_dic) + outputs = func(self, *args, **kwargs) if output_attentions or output_hidden_states: for h in hooks: @@ -1020,7 +1013,6 @@ def hook_fn(module, inp, out): outputs.hidden_states = tuple(collected_hidden_states) if output_attentions: outputs.attentions = tuple(collected_attentions) - if return_dict: outputs = outputs.to_tuple() return outputs From 5abbe87f57fecf13dd08fbd115f6dacb961e0f6e Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:21:01 +0200 Subject: [PATCH 07/51] small nits --- src/transformers/utils/generic.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 0a564b142a8d..2a20e866fdf4 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -981,6 +981,7 @@ def wrapper(self, *args, **kwargs): collected_attentions = [] collected_hidden_states = [] if output_attentions or output_hidden_states: + def output_hidden_and_attention(module, inp, out): if output_hidden_states: collected_hidden_states.append(out[0]) @@ -991,7 +992,6 @@ def output_hidden_and_attention(module, inp, out): if isinstance(layer, GradientCheckpointingLayer): hooks.append(layer.register_forward_hook(output_hidden_and_attention)) - if input_ids ^ inputs_embeds: raise ValueError("You must specify exactly one of input_ids or inputs_embeds.") @@ -1002,7 +1002,6 @@ def output_hidden_and_attention(module, inp, out): if past_key_values is not None and not isinstance(past_key_values, "Cache"): raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") - outputs = func(self, *args, **kwargs) if output_attentions or output_hidden_states: for h in hooks: From 3c791b6f3924185ec7d616832da0bf1330e21b63 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:28:28 +0200 Subject: [PATCH 08/51] fix import --- src/transformers/models/aria/modeling_aria.py | 3 ++- src/transformers/models/bitnet/modeling_bitnet.py | 3 ++- src/transformers/models/cohere/modeling_cohere.py | 3 ++- src/transformers/models/cohere2/modeling_cohere2.py | 3 ++- .../models/deepseek_v3/modeling_deepseek_v3.py | 3 ++- .../models/diffllama/modeling_diffllama.py | 3 ++- src/transformers/models/emu3/modeling_emu3.py | 3 ++- src/transformers/models/gemma/modeling_gemma.py | 3 ++- src/transformers/models/gemma2/modeling_gemma2.py | 3 ++- src/transformers/models/gemma3/modeling_gemma3.py | 10 ++-------- src/transformers/models/glm/modeling_glm.py | 3 ++- src/transformers/models/glm4/modeling_glm4.py | 3 ++- src/transformers/models/gpt_neox/modeling_gpt_neox.py | 3 ++- src/transformers/models/granite/modeling_granite.py | 3 ++- src/transformers/models/helium/modeling_helium.py | 3 ++- src/transformers/models/llama/modeling_llama.py | 3 ++- .../models/moonshine/modeling_moonshine.py | 3 ++- src/transformers/models/olmo/modeling_olmo.py | 3 ++- src/transformers/models/olmo2/modeling_olmo2.py | 3 ++- src/transformers/models/phi/modeling_phi.py | 3 ++- 20 files changed, 40 insertions(+), 27 deletions(-) diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index 55bd10d06116..388902017f24 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -32,7 +32,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs +from ...utils import LossKwargs, auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from ...utils.import_utils import is_torch_available from ..auto import AutoModel from .configuration_aria import AriaConfig, AriaTextConfig diff --git a/src/transformers/models/bitnet/modeling_bitnet.py b/src/transformers/models/bitnet/modeling_bitnet.py index af60de1d8133..780f74e8b146 100644 --- a/src/transformers/models/bitnet/modeling_bitnet.py +++ b/src/transformers/models/bitnet/modeling_bitnet.py @@ -34,7 +34,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs +from ...utils import auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from .configuration_bitnet import BitNetConfig diff --git a/src/transformers/models/cohere/modeling_cohere.py b/src/transformers/models/cohere/modeling_cohere.py index 148a2e8e640d..819fe7f27b93 100644 --- a/src/transformers/models/cohere/modeling_cohere.py +++ b/src/transformers/models/cohere/modeling_cohere.py @@ -42,7 +42,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs +from ...utils import LossKwargs, auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from .configuration_cohere import CohereConfig diff --git a/src/transformers/models/cohere2/modeling_cohere2.py b/src/transformers/models/cohere2/modeling_cohere2.py index 8c675573754b..ca9c4934adda 100644 --- a/src/transformers/models/cohere2/modeling_cohere2.py +++ b/src/transformers/models/cohere2/modeling_cohere2.py @@ -34,8 +34,9 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging from ...utils.deprecation import deprecate_kwarg +from ...utils.generic import check_model_inputs from .configuration_cohere2 import Cohere2Config diff --git a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py index 0884c2ab0530..833d3b0d8b25 100644 --- a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py +++ b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py @@ -22,7 +22,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs +from ...utils import auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from .configuration_deepseek_v3 import DeepseekV3Config diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 8040534544cb..a930e42d7d14 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -48,7 +48,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_diffllama import DiffLlamaConfig diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index ef078dadfbd8..a2d1c77ae36d 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -39,7 +39,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_emu3 import Emu3Config, Emu3TextConfig, Emu3VQVAEConfig diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py index 791e52ba2538..5cb36399fea5 100644 --- a/src/transformers/models/gemma/modeling_gemma.py +++ b/src/transformers/models/gemma/modeling_gemma.py @@ -39,7 +39,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_gemma import GemmaConfig diff --git a/src/transformers/models/gemma2/modeling_gemma2.py b/src/transformers/models/gemma2/modeling_gemma2.py index c21171f5ef3d..0f1f809f35a0 100644 --- a/src/transformers/models/gemma2/modeling_gemma2.py +++ b/src/transformers/models/gemma2/modeling_gemma2.py @@ -39,8 +39,9 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging from ...utils.deprecation import deprecate_kwarg +from ...utils.generic import check_model_inputs from .configuration_gemma2 import Gemma2Config diff --git a/src/transformers/models/gemma3/modeling_gemma3.py b/src/transformers/models/gemma3/modeling_gemma3.py index 403ed9077ee8..54e227d60643 100644 --- a/src/transformers/models/gemma3/modeling_gemma3.py +++ b/src/transformers/models/gemma3/modeling_gemma3.py @@ -38,15 +38,9 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import ( - ModelOutput, - auto_docstring, - can_return_tuple, - check_model_inputs, - is_torchdynamo_compiling, - logging, -) +from ...utils import ModelOutput, auto_docstring, can_return_tuple, is_torchdynamo_compiling, logging from ...utils.deprecation import deprecate_kwarg +from ...utils.generic import check_model_inputs from ..auto import AutoModel from .configuration_gemma3 import Gemma3Config, Gemma3TextConfig diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index 113e0e64b9eb..062120f85155 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -40,7 +40,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_glm import GlmConfig diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index 58c2c9944e35..8b8a64e95988 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -40,7 +40,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_glm4 import Glm4Config diff --git a/src/transformers/models/gpt_neox/modeling_gpt_neox.py b/src/transformers/models/gpt_neox/modeling_gpt_neox.py index 32533cf5cd60..0b96f19befc3 100755 --- a/src/transformers/models/gpt_neox/modeling_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modeling_gpt_neox.py @@ -24,7 +24,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_gpt_neox import GPTNeoXConfig diff --git a/src/transformers/models/granite/modeling_granite.py b/src/transformers/models/granite/modeling_granite.py index 676f21b5f723..73fc1753735f 100644 --- a/src/transformers/models/granite/modeling_granite.py +++ b/src/transformers/models/granite/modeling_granite.py @@ -35,7 +35,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import LossKwargs, auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_granite import GraniteConfig diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index 6fcce60c5908..a961502e9b60 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -40,7 +40,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_helium import HeliumConfig diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index c524b798ffb9..72c8d82fecc5 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -41,7 +41,8 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...pytorch_utils import ALL_LAYERNORM_LAYERS -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_llama import LlamaConfig diff --git a/src/transformers/models/moonshine/modeling_moonshine.py b/src/transformers/models/moonshine/modeling_moonshine.py index 88323c4bb3d5..6dc4fd1e5265 100644 --- a/src/transformers/models/moonshine/modeling_moonshine.py +++ b/src/transformers/models/moonshine/modeling_moonshine.py @@ -41,7 +41,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_moonshine import MoonshineConfig diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py index 9786ddeaff07..e545158622b3 100644 --- a/src/transformers/models/olmo/modeling_olmo.py +++ b/src/transformers/models/olmo/modeling_olmo.py @@ -20,7 +20,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs +from ...utils import auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from .configuration_olmo import OlmoConfig diff --git a/src/transformers/models/olmo2/modeling_olmo2.py b/src/transformers/models/olmo2/modeling_olmo2.py index 958a864a37b9..9a95411eaf8a 100644 --- a/src/transformers/models/olmo2/modeling_olmo2.py +++ b/src/transformers/models/olmo2/modeling_olmo2.py @@ -20,7 +20,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs +from ...utils import auto_docstring, can_return_tuple +from ...utils.generic import check_model_inputs from .configuration_olmo2 import Olmo2Config diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index 8b9b15d6a613..b87f8505d0bc 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -24,7 +24,8 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import auto_docstring, can_return_tuple, check_model_inputs, logging +from ...utils import auto_docstring, can_return_tuple, logging +from ...utils.generic import check_model_inputs from .configuration_phi import PhiConfig From afecb000aa38109dc58f1c64419e034de1867e09 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 11:31:10 +0200 Subject: [PATCH 09/51] small fixes --- src/transformers/utils/generic.py | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 2a20e866fdf4..6ed88b2e1c4f 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -968,7 +968,7 @@ def wrapper(self, *args, **kwargs): kwargs["output_attention"] = kwargs.get("output_attention", self.config.output_attentions) kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) - + kwargs["return_dict"] = True output_attentions = kwargs["output_attention"] output_hidden_states = kwargs["output_hidden_states"] input_ids = kwargs.get("input_ids", None) @@ -992,14 +992,14 @@ def output_hidden_and_attention(module, inp, out): if isinstance(layer, GradientCheckpointingLayer): hooks.append(layer.register_forward_hook(output_hidden_and_attention)) - if input_ids ^ inputs_embeds: + if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds.") if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") kwargs["use_cache"] = False # update it directly in kwargs - if past_key_values is not None and not isinstance(past_key_values, "Cache"): + if past_key_values is not None and not "Cache" in past_key_values.__class__.__name__: raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") outputs = func(self, *args, **kwargs) From 818cef60ea2d09090cac0d43f7d3f54f305a8eeb Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 12:30:39 +0200 Subject: [PATCH 10/51] some small fixes --- .../models/llama/modeling_llama.py | 10 ------- src/transformers/utils/generic.py | 29 ++++++++++--------- 2 files changed, 16 insertions(+), 23 deletions(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index 72c8d82fecc5..c4f2660bcd20 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -469,8 +469,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -497,12 +495,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -510,8 +502,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 6ed88b2e1c4f..b40ef22d0c2a 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -930,7 +930,7 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): - is_requested_to_return_tuple = kwargs.pop("return_dict", True) is False + is_requested_to_return_tuple = kwargs.get("return_dict", True) is False is_configured_to_return_tuple = self.config.use_return_dict is False if hasattr(self, "config") else False # The following allows to convert output to tuple ONLY on top level forward call, @@ -965,18 +965,21 @@ def check_model_inputs(func): @wraps(func) def wrapper(self, *args, **kwargs): - kwargs["output_attention"] = kwargs.get("output_attention", self.config.output_attentions) - kwargs["output_hidden_states"] = kwargs.get("output_hidden_states", self.config.output_hidden_states) - kwargs["use_cache"] = kwargs.get("use_cache", self.config.use_cache) + print(kwargs) + output_attentions = kwargs.get("output_attentions", self.config.output_attentions) + output_hidden_states = kwargs.get("output_hidden_states", self.config.output_hidden_states) + use_cache = kwargs.get("use_cache", self.config.use_cache) + return_dict = kwargs.get("return_dict", self.config.use_return_dict) + + kwargs["output_attentions"] = output_attentions + kwargs["output_hidden_states"] = output_hidden_states + kwargs["use_cache"] = use_cache kwargs["return_dict"] = True - output_attentions = kwargs["output_attention"] - output_hidden_states = kwargs["output_hidden_states"] + input_ids = kwargs.get("input_ids", None) inputs_embeds = kwargs.get("inputs_embeds", None) use_cache = kwargs["use_cache"] past_key_values = kwargs.get("past_key_values", None) - return_dict = kwargs.get("return_dict", self.config.use_return_dict) - hooks = [] collected_attentions = [] collected_hidden_states = [] @@ -985,11 +988,11 @@ def wrapper(self, *args, **kwargs): def output_hidden_and_attention(module, inp, out): if output_hidden_states: collected_hidden_states.append(out[0]) - if output_attentions and len(out) > 1: + if output_attentions: collected_attentions.append(out[1]) - for layer in self.named_modules(): - if isinstance(layer, GradientCheckpointingLayer): + for _, layer in self.named_modules(): + if isinstance(layer, GradientCheckpointingLayer): # TODO a bit slow to iterate over all layers no? hooks.append(layer.register_forward_hook(output_hidden_and_attention)) if (input_ids is None) ^ (inputs_embeds is not None): @@ -999,7 +1002,7 @@ def output_hidden_and_attention(module, inp, out): logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") kwargs["use_cache"] = False # update it directly in kwargs - if past_key_values is not None and not "Cache" in past_key_values.__class__.__name__: + if past_key_values is not None and "Cache" not in past_key_values.__class__.__name__: raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") outputs = func(self, *args, **kwargs) @@ -1012,7 +1015,7 @@ def output_hidden_and_attention(module, inp, out): outputs.hidden_states = tuple(collected_hidden_states) if output_attentions: outputs.attentions = tuple(collected_attentions) - if return_dict: + if not return_dict: outputs = outputs.to_tuple() return outputs From 8f4561778193e42ebfeee3935771d78410afd829 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 12:48:27 +0200 Subject: [PATCH 11/51] simplify --- src/transformers/utils/generic.py | 38 ++++++++++++------------------- 1 file changed, 14 insertions(+), 24 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index b40ef22d0c2a..dac8e68c448e 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -930,28 +930,15 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): - is_requested_to_return_tuple = kwargs.get("return_dict", True) is False - is_configured_to_return_tuple = self.config.use_return_dict is False if hasattr(self, "config") else False - - # The following allows to convert output to tuple ONLY on top level forward call, - # while internal modules of the model will return Output objects - # to be able to use name-based attribute access in modeling code. - - # We will check if we are on top level module, if so, turn off to tuple conversion for all - # underling calls. - is_top_level_module = getattr(self, "_is_top_level_module", True) - if is_configured_to_return_tuple and is_top_level_module: - set_attribute_for_modules(self, "_is_top_level_module", False) - - try: - output = func(self, *args, **kwargs) - if is_requested_to_return_tuple or (is_configured_to_return_tuple and is_top_level_module): - output = output.to_tuple() - finally: - # Remove the flag after the model forward call is finished. - if is_configured_to_return_tuple and is_top_level_module: - del_attribute_from_modules(self, "_is_top_level_module") - + return_dict = kwargs.get("return_dict", self.config.use_return_dict) + kwargs["return_dict"] = False # always set return_dict to True for the function call + output = func(self, *args, **kwargs) + if not return_dict: + import inspect + frame = inspect.stack() # 3 because we're interested in the caller (not the function itself) + caller_obj = frame[3][0].f_locals.get('self', None) + if caller_obj.__class__ == self.__class__: + outputs = outputs.to_tuple() return output return wrapper @@ -969,7 +956,7 @@ def wrapper(self, *args, **kwargs): output_attentions = kwargs.get("output_attentions", self.config.output_attentions) output_hidden_states = kwargs.get("output_hidden_states", self.config.output_hidden_states) use_cache = kwargs.get("use_cache", self.config.use_cache) - return_dict = kwargs.get("return_dict", self.config.use_return_dict) + return_dict = kwargs.pop("return_dict", self.config.use_return_dict) kwargs["output_attentions"] = output_attentions kwargs["output_hidden_states"] = output_hidden_states @@ -1015,7 +1002,10 @@ def output_hidden_and_attention(module, inp, out): outputs.hidden_states = tuple(collected_hidden_states) if output_attentions: outputs.attentions = tuple(collected_attentions) - if not return_dict: + import inspect + frame = inspect.stack() # 2 because we're interested in the caller (not the function itself) + caller_obj = frame[3][0].f_locals.get('self', None) + if caller_obj.__class__ == self.__class__ and not return_dict: outputs = outputs.to_tuple() return outputs From da51096bca497556b995fb4a9cf94c7d561ad088 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 13:10:20 +0200 Subject: [PATCH 12/51] cache the caller for now --- src/transformers/utils/generic.py | 10 ++++++---- 1 file changed, 6 insertions(+), 4 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index dac8e68c448e..a7acd3ecc0eb 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -934,10 +934,12 @@ def wrapper(self, *args, **kwargs): kwargs["return_dict"] = False # always set return_dict to True for the function call output = func(self, *args, **kwargs) if not return_dict: - import inspect - frame = inspect.stack() # 3 because we're interested in the caller (not the function itself) - caller_obj = frame[3][0].f_locals.get('self', None) - if caller_obj.__class__ == self.__class__: + if not hasattr(self, "_caller_obj"): + import inspect + frame = inspect.stack() # 3 because we're interested in the caller (not the function itself) + self._caller_obj = frame[3][0].f_locals.get('self', None) + + if self._caller_obj.__class__ == self.__class__: outputs = outputs.to_tuple() return output From 14289072e57a7032f28a2557593c80c012b53b24 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Mon, 26 May 2025 14:01:56 +0200 Subject: [PATCH 13/51] done for the day gotta jumb on other stuffs --- src/transformers/utils/generic.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index a7acd3ecc0eb..bde6fe92e9ac 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -935,7 +935,7 @@ def wrapper(self, *args, **kwargs): output = func(self, *args, **kwargs) if not return_dict: if not hasattr(self, "_caller_obj"): - import inspect + frame = inspect.stack() # 3 because we're interested in the caller (not the function itself) self._caller_obj = frame[3][0].f_locals.get('self', None) From 081ebcc01c238cd239daac9b5337fba11add70f8 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:29:32 +0200 Subject: [PATCH 14/51] mega simplify --- src/transformers/utils/generic.py | 20 +++++--------------- 1 file changed, 5 insertions(+), 15 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index bde6fe92e9ac..e27c1850fb29 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -931,18 +931,11 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): return_dict = kwargs.get("return_dict", self.config.use_return_dict) - kwargs["return_dict"] = False # always set return_dict to True for the function call + kwargs["return_dict"] = True output = func(self, *args, **kwargs) - if not return_dict: - if not hasattr(self, "_caller_obj"): - - frame = inspect.stack() # 3 because we're interested in the caller (not the function itself) - self._caller_obj = frame[3][0].f_locals.get('self', None) - - if self._caller_obj.__class__ == self.__class__: - outputs = outputs.to_tuple() + if user_return_dict is False: + output = output.to_tuple() return output - return wrapper @@ -1004,11 +997,8 @@ def output_hidden_and_attention(module, inp, out): outputs.hidden_states = tuple(collected_hidden_states) if output_attentions: outputs.attentions = tuple(collected_attentions) - import inspect - frame = inspect.stack() # 2 because we're interested in the caller (not the function itself) - caller_obj = frame[3][0].f_locals.get('self', None) - if caller_obj.__class__ == self.__class__ and not return_dict: - outputs = outputs.to_tuple() + if return_dict is False: + output = output.to_tuple() return outputs return wrapper From b5c6501f7cbccfbf5438387deaf608d70aa5d133 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:30:00 +0200 Subject: [PATCH 15/51] much simpler --- src/transformers/utils/generic.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index e27c1850fb29..d5c8881c0555 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -933,9 +933,10 @@ def wrapper(self, *args, **kwargs): return_dict = kwargs.get("return_dict", self.config.use_return_dict) kwargs["return_dict"] = True output = func(self, *args, **kwargs) - if user_return_dict is False: + if return_dict is False: output = output.to_tuple() return output + return wrapper @@ -998,7 +999,7 @@ def output_hidden_and_attention(module, inp, out): if output_attentions: outputs.attentions = tuple(collected_attentions) if return_dict is False: - output = output.to_tuple() + outputs = outputs.to_tuple() return outputs return wrapper From 5b95608bf1c2da6fc4b73547142fb0c038d638ac Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:37:25 +0200 Subject: [PATCH 16/51] noice! --- src/transformers/models/aria/modeling_aria.py | 10 ---------- src/transformers/models/bamba/modeling_bamba.py | 8 -------- src/transformers/models/bitnet/modeling_bitnet.py | 10 ---------- .../models/deepseek_v3/modeling_deepseek_v3.py | 10 ---------- .../models/diffllama/modeling_diffllama.py | 10 ---------- src/transformers/models/emu3/modeling_emu3.py | 10 ---------- src/transformers/models/gemma/modeling_gemma.py | 10 ---------- src/transformers/models/glm/modeling_glm.py | 10 ---------- src/transformers/models/glm4/modeling_glm4.py | 10 ---------- src/transformers/models/helium/modeling_helium.py | 10 ---------- src/transformers/models/mistral/modeling_mistral.py | 10 ---------- src/transformers/models/olmo/modeling_olmo.py | 10 ---------- src/transformers/models/olmo2/modeling_olmo2.py | 10 ---------- src/transformers/models/phi/modeling_phi.py | 10 ---------- src/transformers/models/phi3/modeling_phi3.py | 10 ---------- src/transformers/models/qwen2/modeling_qwen2.py | 10 ---------- src/transformers/models/qwen3/modeling_qwen3.py | 10 ---------- .../models/starcoder2/modeling_starcoder2.py | 10 ---------- 18 files changed, 178 deletions(-) diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index 388902017f24..43b116a520d4 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -848,8 +848,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -876,12 +874,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -889,8 +881,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/bamba/modeling_bamba.py b/src/transformers/models/bamba/modeling_bamba.py index 95af1f251369..3bdeae6b7593 100644 --- a/src/transformers/models/bamba/modeling_bamba.py +++ b/src/transformers/models/bamba/modeling_bamba.py @@ -1392,12 +1392,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -1405,8 +1399,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/bitnet/modeling_bitnet.py b/src/transformers/models/bitnet/modeling_bitnet.py index 780f74e8b146..947217509425 100644 --- a/src/transformers/models/bitnet/modeling_bitnet.py +++ b/src/transformers/models/bitnet/modeling_bitnet.py @@ -460,8 +460,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -488,12 +486,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "User: Hey, are you conscious? Can you talk to me?Assistant: No, I'm not conscious. I'm an artificial intelligence designed to assist with information and tasks. How can I help you today?" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -501,8 +493,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py index 833d3b0d8b25..691e9c6f9ad2 100644 --- a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py +++ b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py @@ -650,8 +650,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -678,12 +676,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -691,8 +683,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index a930e42d7d14..5dd962a7f480 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -735,8 +735,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -763,12 +761,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -776,8 +768,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index a2d1c77ae36d..3b8bd35a2d0d 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -1323,8 +1323,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -1351,12 +1349,6 @@ def forward( >>> generated_ids = model.generate(**inputs, max_new_tokens=100, do_sample=False) >>> processor.batch_decode(generated_ids, skip_special_tokens=True)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -1364,8 +1356,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py index 5cb36399fea5..9d5c0c129d56 100644 --- a/src/transformers/models/gemma/modeling_gemma.py +++ b/src/transformers/models/gemma/modeling_gemma.py @@ -508,8 +508,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -536,12 +534,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -549,8 +541,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index 062120f85155..f06f3dc288ec 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -480,8 +480,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -508,12 +506,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -521,8 +513,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index 8b8a64e95988..d860071a621c 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -493,8 +493,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -521,12 +519,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -534,8 +526,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index a961502e9b60..1fb115651d09 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -465,8 +465,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -493,12 +491,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -506,8 +498,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py index f3c1a2837546..7a98a39c792f 100644 --- a/src/transformers/models/mistral/modeling_mistral.py +++ b/src/transformers/models/mistral/modeling_mistral.py @@ -483,8 +483,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -511,12 +509,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -524,8 +516,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py index e545158622b3..01c7eeac60b5 100644 --- a/src/transformers/models/olmo/modeling_olmo.py +++ b/src/transformers/models/olmo/modeling_olmo.py @@ -441,8 +441,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -469,12 +467,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -482,8 +474,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/olmo2/modeling_olmo2.py b/src/transformers/models/olmo2/modeling_olmo2.py index 9a95411eaf8a..591d392d66fd 100644 --- a/src/transformers/models/olmo2/modeling_olmo2.py +++ b/src/transformers/models/olmo2/modeling_olmo2.py @@ -452,8 +452,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -480,12 +478,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -493,8 +485,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index b87f8505d0bc..3fe8445fb985 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -499,8 +499,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -527,12 +525,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -540,8 +532,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index 6a94e0fb3d1f..0635a56f9cd8 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -543,8 +543,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -571,12 +569,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -584,8 +576,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py index d1dd12916a13..b9bfb8f2087f 100644 --- a/src/transformers/models/qwen2/modeling_qwen2.py +++ b/src/transformers/models/qwen2/modeling_qwen2.py @@ -497,8 +497,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -525,12 +523,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -538,8 +530,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 47d045c216a2..57208c9071da 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -526,8 +526,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -554,12 +552,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -567,8 +559,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py index 539db55ffbed..e6940e170fb2 100644 --- a/src/transformers/models/starcoder2/modeling_starcoder2.py +++ b/src/transformers/models/starcoder2/modeling_starcoder2.py @@ -487,8 +487,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[FlashAttentionKwargs], @@ -515,12 +513,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, attention_mask=attention_mask, @@ -528,8 +520,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) From b76a44d2e22b677380a6d75f21dd9cd9a9961d9f Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:38:02 +0200 Subject: [PATCH 17/51] this one will be for another PR, it's a bit annoying to handle right now --- src/transformers/models/llama/modeling_llama.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index c4f2660bcd20..397dede7a823 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -41,7 +41,7 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...pytorch_utils import ALL_LAYERNORM_LAYERS -from ...utils import auto_docstring, can_return_tuple, logging +from ...utils import auto_docstring, can_return_tuple, logging, LossKwargs from ...utils.generic import check_model_inputs from .configuration_llama import LlamaConfig @@ -425,6 +425,9 @@ def forward( ) +class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... + + @auto_docstring class LlamaForCausalLM(LlamaPreTrainedModel, GenerationMixin): _tied_weights_keys = ["lm_head.weight"] From d0035f754f3fcbe821501dff66ba7a7d094fe7a4 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:39:22 +0200 Subject: [PATCH 18/51] weird --- src/transformers/models/csm/modeling_csm.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/transformers/models/csm/modeling_csm.py b/src/transformers/models/csm/modeling_csm.py index 244a8bb36bea..2aa69f8440f1 100644 --- a/src/transformers/models/csm/modeling_csm.py +++ b/src/transformers/models/csm/modeling_csm.py @@ -36,7 +36,7 @@ from ...modeling_rope_utils import ROPE_INIT_FUNCTIONS, dynamic_rope_update from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack -from ...utils import ModelOutput, auto_docstring, can_return_tuple, logging +from ...utils import LossKwargs, ModelOutput, auto_docstring, can_return_tuple, logging from ..auto import AutoModel from .configuration_csm import CsmConfig, CsmDepthDecoderConfig from .generation_csm import CsmGenerationMixin @@ -574,6 +574,9 @@ def forward(self, hidden_states, cache_position=None): return hidden_states +class KwargsForCausalLM(FlashAttentionKwargs, LossKwargs): ... + + @auto_docstring( custom_intro=""" The CsmDepthDecoder Model transformer, with a [`CsmCodebooksHead`] on top, From 91fc267959dd75e4628f2fd26e01c762c9c0f196 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:46:21 +0200 Subject: [PATCH 19/51] fix typing? --- src/transformers/utils/generic.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index d5c8881c0555..3632a9213019 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -31,7 +31,6 @@ import numpy as np from packaging import version -from ..modeling_layers import GradientCheckpointingLayer from ..utils import logging from .import_utils import ( get_torch_version, @@ -48,6 +47,7 @@ if is_torch_available(): # required for @can_return_tuple decorator to work with torchdynamo import torch # noqa: F401 + from ..modeling_layers import GradientCheckpointingLayer class cached_property(property): From 949f149883c5c1801f5768903e44647ade77b959 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 11:48:24 +0200 Subject: [PATCH 20/51] fixup --- src/transformers/models/llama/modeling_llama.py | 2 +- src/transformers/utils/generic.py | 1 + 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index 397dede7a823..a60e2bbc8d84 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -41,7 +41,7 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel from ...processing_utils import Unpack from ...pytorch_utils import ALL_LAYERNORM_LAYERS -from ...utils import auto_docstring, can_return_tuple, logging, LossKwargs +from ...utils import LossKwargs, auto_docstring, can_return_tuple, logging from ...utils.generic import check_model_inputs from .configuration_llama import LlamaConfig diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 3632a9213019..3ee572ee2717 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -47,6 +47,7 @@ if is_torch_available(): # required for @can_return_tuple decorator to work with torchdynamo import torch # noqa: F401 + from ..modeling_layers import GradientCheckpointingLayer From 4affc579d83ab2e826f84f4055c33a12bd5fb8f8 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 12:07:58 +0200 Subject: [PATCH 21/51] fix the arg issues --- src/transformers/utils/generic.py | 29 ++++++++++++++++++++--------- 1 file changed, 20 insertions(+), 9 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 3ee572ee2717..fa7213722280 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -931,10 +931,12 @@ def can_return_tuple(func): @wraps(func) def wrapper(self, *args, **kwargs): - return_dict = kwargs.get("return_dict", self.config.use_return_dict) - kwargs["return_dict"] = True + if "return_dict" in kwargs: + return_dict = kwargs.get("return_dict", self.config.use_return_dict) + kwargs["return_dict"] = True output = func(self, *args, **kwargs) - if return_dict is False: + + if "return_dict" in kwargs and return_dict is False: output = output.to_tuple() return output @@ -955,13 +957,22 @@ def wrapper(self, *args, **kwargs): use_cache = kwargs.get("use_cache", self.config.use_cache) return_dict = kwargs.pop("return_dict", self.config.use_return_dict) - kwargs["output_attentions"] = output_attentions - kwargs["output_hidden_states"] = output_hidden_states - kwargs["use_cache"] = use_cache - kwargs["return_dict"] = True + # Safely extract common config-backed flags + kwargs.setdefault("output_attentions", output_attentions) + kwargs.setdefault("output_hidden_states", output_hidden_states) + kwargs.setdefault("use_cache", use_cache) + kwargs["return_dict"] = kwargs.pop("return_dict", return_dict) + + # Use inspect to bind args/kwargs to parameter names + sig = inspect.signature(func) + bound = sig.bind_partial(self, *args, **kwargs) + bound.apply_defaults() + all_args = bound.arguments + + # Extract input_ids and inputs_embeds more robustly + input_ids = all_args.get("input_ids", None) + inputs_embeds = all_args.get("inputs_embeds", None) - input_ids = kwargs.get("input_ids", None) - inputs_embeds = kwargs.get("inputs_embeds", None) use_cache = kwargs["use_cache"] past_key_values = kwargs.get("past_key_values", None) hooks = [] From 4e35168f460fb5f09c0a70422343af28e89ec92b Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 15:03:15 +0200 Subject: [PATCH 22/51] ? --- src/transformers/utils/generic.py | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index fa7213722280..fff560178285 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -1007,9 +1007,13 @@ def output_hidden_and_attention(module, inp, out): if output_hidden_states: collected_hidden_states.append(outputs.last_hidden_state) - outputs.hidden_states = tuple(collected_hidden_states) + outputs["hidden_states"] = tuple(collected_hidden_states) + else: + outputs["hidden_states"] = None if output_attentions: - outputs.attentions = tuple(collected_attentions) + outputs["attention"] = tuple(collected_attentions) + else: + outputs["attentions"] = None if return_dict is False: outputs = outputs.to_tuple() return outputs From 71838e4a8f6e554e2eec6a263ec1f6b29cfcfd52 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 15:21:49 +0200 Subject: [PATCH 23/51] lol one single nit --- examples/modular-transformers/modeling_dummy.py | 4 ++-- examples/modular-transformers/modeling_multimodal1.py | 4 ++-- src/transformers/models/aria/modeling_aria.py | 4 ++-- src/transformers/models/biogpt/modeling_biogpt.py | 6 +++--- src/transformers/models/biogpt/modular_biogpt.py | 6 +++--- src/transformers/models/bitnet/modeling_bitnet.py | 4 ++-- src/transformers/models/cohere/modeling_cohere.py | 4 ++-- src/transformers/models/cohere2/modeling_cohere2.py | 4 ++-- src/transformers/models/cohere2/modular_cohere2.py | 4 ++-- src/transformers/models/csm/modeling_csm.py | 6 +++--- src/transformers/models/csm/modular_csm.py | 2 +- src/transformers/models/deepseek_v3/modeling_deepseek_v3.py | 4 ++-- src/transformers/models/diffllama/modeling_diffllama.py | 4 ++-- src/transformers/models/emu3/modeling_emu3.py | 4 ++-- src/transformers/models/gemma2/modeling_gemma2.py | 6 +++--- src/transformers/models/gemma2/modular_gemma2.py | 6 +++--- src/transformers/models/gemma3/modeling_gemma3.py | 6 +++--- src/transformers/models/gemma3/modular_gemma3.py | 6 +++--- src/transformers/models/glm/modeling_glm.py | 4 ++-- src/transformers/models/glm4/modeling_glm4.py | 4 ++-- src/transformers/models/gpt_neox/modeling_gpt_neox.py | 4 ++-- src/transformers/models/gpt_neox/modular_gpt_neox.py | 4 ++-- src/transformers/models/granite/modeling_granite.py | 4 ++-- src/transformers/models/granite/modular_granite.py | 4 ++-- src/transformers/models/helium/modeling_helium.py | 4 ++-- src/transformers/models/llama/modeling_llama.py | 4 ++-- src/transformers/models/llama4/modeling_llama4.py | 2 +- src/transformers/models/mistral/modeling_mistral.py | 4 ++-- src/transformers/models/mistral/modular_mistral.py | 4 ++-- src/transformers/models/mixtral/modeling_mixtral.py | 6 +++--- src/transformers/models/mixtral/modular_mixtral.py | 6 +++--- src/transformers/models/moonshine/modeling_moonshine.py | 6 +++--- src/transformers/models/moonshine/modular_moonshine.py | 6 +++--- src/transformers/models/olmo/modeling_olmo.py | 4 ++-- src/transformers/models/olmo2/modeling_olmo2.py | 4 ++-- src/transformers/models/phi/modeling_phi.py | 6 +++--- src/transformers/models/phi/modular_phi.py | 6 +++--- src/transformers/models/phi3/modeling_phi3.py | 4 ++-- src/transformers/models/qwen2/modeling_qwen2.py | 4 ++-- src/transformers/models/qwen2/modular_qwen2.py | 4 ++-- src/transformers/models/qwen3/modeling_qwen3.py | 4 ++-- src/transformers/models/qwen3_moe/modeling_qwen3_moe.py | 6 +++--- src/transformers/models/starcoder2/modeling_starcoder2.py | 4 ++-- src/transformers/models/starcoder2/modular_starcoder2.py | 4 ++-- src/transformers/utils/generic.py | 2 +- tests/utils/test_auto_docstring.py | 6 +++--- 46 files changed, 104 insertions(+), 104 deletions(-) diff --git a/examples/modular-transformers/modeling_dummy.py b/examples/modular-transformers/modeling_dummy.py index 0fe4ae497b49..ddcf808d9935 100644 --- a/examples/modular-transformers/modeling_dummy.py +++ b/examples/modular-transformers/modeling_dummy.py @@ -477,7 +477,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -539,7 +539,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/examples/modular-transformers/modeling_multimodal1.py b/examples/modular-transformers/modeling_multimodal1.py index ec1a3346c9ba..c16135b6bc42 100644 --- a/examples/modular-transformers/modeling_multimodal1.py +++ b/examples/modular-transformers/modeling_multimodal1.py @@ -477,7 +477,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -539,7 +539,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index 43b116a520d4..24894b817bed 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -754,7 +754,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -790,7 +790,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/biogpt/modeling_biogpt.py b/src/transformers/models/biogpt/modeling_biogpt.py index a1fba008841f..440a40f9b6f2 100755 --- a/src/transformers/models/biogpt/modeling_biogpt.py +++ b/src/transformers/models/biogpt/modeling_biogpt.py @@ -315,7 +315,7 @@ def forward( output_attentions=output_attentions, position_ids=position_ids, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training) hidden_states = residual + hidden_states @@ -648,7 +648,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, head_mask[idx] if head_mask is not None else None, @@ -668,7 +668,7 @@ def forward( use_cache=use_cache, position_ids=position_ids, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/biogpt/modular_biogpt.py b/src/transformers/models/biogpt/modular_biogpt.py index d639f44ffec5..f37e15e0f079 100644 --- a/src/transformers/models/biogpt/modular_biogpt.py +++ b/src/transformers/models/biogpt/modular_biogpt.py @@ -142,7 +142,7 @@ def forward( output_attentions=output_attentions, position_ids=position_ids, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training) hidden_states = residual + hidden_states @@ -475,7 +475,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, head_mask[idx] if head_mask is not None else None, @@ -495,7 +495,7 @@ def forward( use_cache=use_cache, position_ids=position_ids, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/bitnet/modeling_bitnet.py b/src/transformers/models/bitnet/modeling_bitnet.py index 947217509425..dd6575adc643 100644 --- a/src/transformers/models/bitnet/modeling_bitnet.py +++ b/src/transformers/models/bitnet/modeling_bitnet.py @@ -368,7 +368,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -404,7 +404,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/cohere/modeling_cohere.py b/src/transformers/models/cohere/modeling_cohere.py index 819fe7f27b93..444e026ef768 100644 --- a/src/transformers/models/cohere/modeling_cohere.py +++ b/src/transformers/models/cohere/modeling_cohere.py @@ -410,7 +410,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -446,7 +446,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/cohere2/modeling_cohere2.py b/src/transformers/models/cohere2/modeling_cohere2.py index ca9c4934adda..48b90383f824 100644 --- a/src/transformers/models/cohere2/modeling_cohere2.py +++ b/src/transformers/models/cohere2/modeling_cohere2.py @@ -394,7 +394,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -462,7 +462,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/cohere2/modular_cohere2.py b/src/transformers/models/cohere2/modular_cohere2.py index b32e4d94dd5c..f14ffc85fdfa 100644 --- a/src/transformers/models/cohere2/modular_cohere2.py +++ b/src/transformers/models/cohere2/modular_cohere2.py @@ -415,7 +415,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -483,7 +483,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/csm/modeling_csm.py b/src/transformers/models/csm/modeling_csm.py index 2aa69f8440f1..67c8365b5881 100644 --- a/src/transformers/models/csm/modeling_csm.py +++ b/src/transformers/models/csm/modeling_csm.py @@ -529,7 +529,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] @@ -759,7 +759,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length, num_codebooks) or (batch_size, sequence_length)`): @@ -807,7 +807,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/csm/modular_csm.py b/src/transformers/models/csm/modular_csm.py index 4bfc68504067..7f882ca2c0ec 100644 --- a/src/transformers/models/csm/modular_csm.py +++ b/src/transformers/models/csm/modular_csm.py @@ -272,7 +272,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py index 691e9c6f9ad2..096aa4bfdde0 100644 --- a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py +++ b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py @@ -558,7 +558,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -594,7 +594,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 5dd962a7f480..4a44f081dac7 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -643,7 +643,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -679,7 +679,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index 3b8bd35a2d0d..c0c01b34a17d 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -1230,7 +1230,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -1266,7 +1266,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/gemma2/modeling_gemma2.py b/src/transformers/models/gemma2/modeling_gemma2.py index 0f1f809f35a0..9cda62d400e1 100644 --- a/src/transformers/models/gemma2/modeling_gemma2.py +++ b/src/transformers/models/gemma2/modeling_gemma2.py @@ -399,7 +399,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -469,7 +469,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, position_embeddings, causal_mask_mapping[decoder_layer.attention_type], @@ -489,7 +489,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/gemma2/modular_gemma2.py b/src/transformers/models/gemma2/modular_gemma2.py index 9890711d3c1f..77c38ad5117b 100644 --- a/src/transformers/models/gemma2/modular_gemma2.py +++ b/src/transformers/models/gemma2/modular_gemma2.py @@ -381,7 +381,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -451,7 +451,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, position_embeddings, causal_mask_mapping[decoder_layer.attention_type], @@ -471,7 +471,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/gemma3/modeling_gemma3.py b/src/transformers/models/gemma3/modeling_gemma3.py index 54e227d60643..c42504e8ba1e 100644 --- a/src/transformers/models/gemma3/modeling_gemma3.py +++ b/src/transformers/models/gemma3/modeling_gemma3.py @@ -517,7 +517,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -584,7 +584,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, position_embeddings_global, position_embeddings_local, @@ -606,7 +606,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/gemma3/modular_gemma3.py b/src/transformers/models/gemma3/modular_gemma3.py index d93d53c8e93c..3ce4bb4813c4 100644 --- a/src/transformers/models/gemma3/modular_gemma3.py +++ b/src/transformers/models/gemma3/modular_gemma3.py @@ -567,7 +567,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -634,7 +634,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, position_embeddings_global, position_embeddings_local, @@ -656,7 +656,7 @@ def forward( output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index f06f3dc288ec..0656ac4e827a 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -388,7 +388,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -424,7 +424,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index d860071a621c..7524006a293d 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -401,7 +401,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -437,7 +437,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/gpt_neox/modeling_gpt_neox.py b/src/transformers/models/gpt_neox/modeling_gpt_neox.py index 0b96f19befc3..8e00022b0b7b 100755 --- a/src/transformers/models/gpt_neox/modeling_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modeling_gpt_neox.py @@ -352,7 +352,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -440,7 +440,7 @@ def forward( output_attentions=output_attentions, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = outputs[0] diff --git a/src/transformers/models/gpt_neox/modular_gpt_neox.py b/src/transformers/models/gpt_neox/modular_gpt_neox.py index 4922a4e3b4c3..095188d21009 100644 --- a/src/transformers/models/gpt_neox/modular_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modular_gpt_neox.py @@ -298,7 +298,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -386,7 +386,7 @@ def forward( output_attentions=output_attentions, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = outputs[0] diff --git a/src/transformers/models/granite/modeling_granite.py b/src/transformers/models/granite/modeling_granite.py index 73fc1753735f..634032d7a236 100644 --- a/src/transformers/models/granite/modeling_granite.py +++ b/src/transformers/models/granite/modeling_granite.py @@ -400,7 +400,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -464,7 +464,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/granite/modular_granite.py b/src/transformers/models/granite/modular_granite.py index 0432b9e2a5c2..b54d7eec6b00 100644 --- a/src/transformers/models/granite/modular_granite.py +++ b/src/transformers/models/granite/modular_granite.py @@ -141,7 +141,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -205,7 +205,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index 1fb115651d09..f164ac5a2ddc 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -373,7 +373,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -409,7 +409,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index a60e2bbc8d84..c421abf04c74 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -377,7 +377,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -413,7 +413,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/llama4/modeling_llama4.py b/src/transformers/models/llama4/modeling_llama4.py index 38b6fde10372..9cdcc07ad960 100644 --- a/src/transformers/models/llama4/modeling_llama4.py +++ b/src/transformers/models/llama4/modeling_llama4.py @@ -594,7 +594,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=freq_cis, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py index 7a98a39c792f..c6df5f22af3c 100644 --- a/src/transformers/models/mistral/modeling_mistral.py +++ b/src/transformers/models/mistral/modeling_mistral.py @@ -350,7 +350,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -417,7 +417,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/mistral/modular_mistral.py b/src/transformers/models/mistral/modular_mistral.py index b28ba006e923..0e7238fdef52 100644 --- a/src/transformers/models/mistral/modular_mistral.py +++ b/src/transformers/models/mistral/modular_mistral.py @@ -119,7 +119,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -186,7 +186,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/mixtral/modeling_mixtral.py b/src/transformers/models/mixtral/modeling_mixtral.py index 26007b7b18ac..4884de54d076 100644 --- a/src/transformers/models/mixtral/modeling_mixtral.py +++ b/src/transformers/models/mixtral/modeling_mixtral.py @@ -477,7 +477,7 @@ def forward( output_hidden_states: Optional[bool] = None, output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_router_logits = ( @@ -537,7 +537,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, position_ids, @@ -559,7 +559,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/mixtral/modular_mixtral.py b/src/transformers/models/mixtral/modular_mixtral.py index bfc78597cfe7..221aeb380a9d 100644 --- a/src/transformers/models/mixtral/modular_mixtral.py +++ b/src/transformers/models/mixtral/modular_mixtral.py @@ -328,7 +328,7 @@ def forward( output_hidden_states: Optional[bool] = None, output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_router_logits = ( @@ -388,7 +388,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, position_ids, @@ -410,7 +410,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/moonshine/modeling_moonshine.py b/src/transformers/models/moonshine/modeling_moonshine.py index 6dc4fd1e5265..c8b58869545c 100644 --- a/src/transformers/models/moonshine/modeling_moonshine.py +++ b/src/transformers/models/moonshine/modeling_moonshine.py @@ -552,7 +552,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" Args: @@ -627,7 +627,7 @@ def forward( position_ids=position_ids, output_attentions=output_attentions, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] @@ -786,7 +786,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/moonshine/modular_moonshine.py b/src/transformers/models/moonshine/modular_moonshine.py index 4ee7cd81f772..ea82ce3af3b3 100644 --- a/src/transformers/models/moonshine/modular_moonshine.py +++ b/src/transformers/models/moonshine/modular_moonshine.py @@ -580,7 +580,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" Args: @@ -655,7 +655,7 @@ def forward( position_ids=position_ids, output_attentions=output_attentions, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] @@ -796,7 +796,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py index 01c7eeac60b5..4225ba00b97f 100644 --- a/src/transformers/models/olmo/modeling_olmo.py +++ b/src/transformers/models/olmo/modeling_olmo.py @@ -349,7 +349,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -385,7 +385,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/olmo2/modeling_olmo2.py b/src/transformers/models/olmo2/modeling_olmo2.py index 591d392d66fd..11598c1ad872 100644 --- a/src/transformers/models/olmo2/modeling_olmo2.py +++ b/src/transformers/models/olmo2/modeling_olmo2.py @@ -360,7 +360,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, cache_position: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -396,7 +396,7 @@ def forward( past_key_value=past_key_values, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index 3fe8445fb985..2a3a005d8050 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -357,7 +357,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -413,7 +413,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, position_ids, @@ -433,7 +433,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/phi/modular_phi.py b/src/transformers/models/phi/modular_phi.py index c515e13e7231..762318773664 100644 --- a/src/transformers/models/phi/modular_phi.py +++ b/src/transformers/models/phi/modular_phi.py @@ -207,7 +207,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -263,7 +263,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, position_ids, @@ -283,7 +283,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index 0635a56f9cd8..47833949c96b 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -410,7 +410,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -477,7 +477,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py index b9bfb8f2087f..058e7eb46f4b 100644 --- a/src/transformers/models/qwen2/modeling_qwen2.py +++ b/src/transformers/models/qwen2/modeling_qwen2.py @@ -355,7 +355,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -431,7 +431,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/qwen2/modular_qwen2.py b/src/transformers/models/qwen2/modular_qwen2.py index f73aedc2c07e..90e22f4507c1 100644 --- a/src/transformers/models/qwen2/modular_qwen2.py +++ b/src/transformers/models/qwen2/modular_qwen2.py @@ -122,7 +122,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -198,7 +198,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 57208c9071da..517cf1c5cc3d 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -381,7 +381,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -457,7 +457,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py index 67f21d1b8368..6a084d2022cd 100644 --- a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py +++ b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py @@ -483,7 +483,7 @@ def forward( output_hidden_states: Optional[bool] = None, output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_router_logits = ( @@ -543,7 +543,7 @@ def forward( if self.gradient_checkpointing and self.training: layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **flash_attn_kwargs), + partial(decoder_layer.__call__, **kwargs), hidden_states, causal_mask, position_ids, @@ -565,7 +565,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py index e6940e170fb2..a662bc78eb8f 100644 --- a/src/transformers/models/starcoder2/modeling_starcoder2.py +++ b/src/transformers/models/starcoder2/modeling_starcoder2.py @@ -355,7 +355,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -421,7 +421,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/models/starcoder2/modular_starcoder2.py b/src/transformers/models/starcoder2/modular_starcoder2.py index 722674371331..517695595c54 100644 --- a/src/transformers/models/starcoder2/modular_starcoder2.py +++ b/src/transformers/models/starcoder2/modular_starcoder2.py @@ -175,7 +175,7 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **flash_attn_kwargs: Unpack[FlashAttentionKwargs], + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -241,7 +241,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **flash_attn_kwargs, + **kwargs, ) hidden_states = layer_outputs[0] diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index fff560178285..3f1937e233fc 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -1011,7 +1011,7 @@ def output_hidden_and_attention(module, inp, out): else: outputs["hidden_states"] = None if output_attentions: - outputs["attention"] = tuple(collected_attentions) + outputs["attentions"] = tuple(collected_attentions) else: outputs["attentions"] = None if return_dict is False: diff --git a/tests/utils/test_auto_docstring.py b/tests/utils/test_auto_docstring.py index 11ac05f3b178..22b0520453ef 100644 --- a/tests/utils/test_auto_docstring.py +++ b/tests/utils/test_auto_docstring.py @@ -48,7 +48,7 @@ class AutoDocstringTest(unittest.TestCase): # COOL_CLASS_DOC = """ # Args: # input_ids (some): - # flash_attn_kwargs (FlashAttentionKwrargs): + # kwargs (FlashAttentionKwrargs): # parameters that are completely optional and that should be passed. # another_warg (something): should pass # and_another_on (this time): @@ -66,10 +66,10 @@ class AutoDocstringTest(unittest.TestCase): # @auto_docstring # class MyModel: # @auto_docstring - # def __init__(input_ids, flash_attn_kwargs=None, another_warg=True, and_another_on=1): + # def __init__(input_ids, kwargs=None, another_warg=True, and_another_on=1): # r""" # Args: - # flash_attn_kwargs (FlashAttentionKwrargs): + # kwargs (FlashAttentionKwrargs): # parameters that are completely optional and that should be passed. # another_warg (something): should pass # and_another_on (this time): From 46b1b90b47855bc1f59a8884431ea7eaeffab5ac Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 15:29:09 +0200 Subject: [PATCH 24/51] update --- src/transformers/utils/generic.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 3f1937e233fc..a5a8c67bc124 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -1008,11 +1008,11 @@ def output_hidden_and_attention(module, inp, out): if output_hidden_states: collected_hidden_states.append(outputs.last_hidden_state) outputs["hidden_states"] = tuple(collected_hidden_states) - else: + elif not return_dict: outputs["hidden_states"] = None if output_attentions: outputs["attentions"] = tuple(collected_attentions) - else: + elif not return_dict: outputs["attentions"] = None if return_dict is False: outputs = outputs.to_tuple() From b7ae7d92a7b92ccbd7271350f450195d5370c773 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 15:52:31 +0200 Subject: [PATCH 25/51] keep one check explicit --- src/transformers/models/aria/modeling_aria.py | 3 +++ .../models/bitnet/modeling_bitnet.py | 3 +++ .../models/cohere/modeling_cohere.py | 3 +++ src/transformers/models/csm/modeling_csm.py | 3 +++ .../models/deepseek_v3/modeling_deepseek_v3.py | 3 +++ .../models/diffllama/modeling_diffllama.py | 3 +++ src/transformers/models/emu3/modeling_emu3.py | 3 +++ src/transformers/models/glm/modeling_glm.py | 3 +++ src/transformers/models/glm4/modeling_glm4.py | 3 +++ .../models/helium/modeling_helium.py | 3 +++ .../models/llama/modeling_llama.py | 3 +++ src/transformers/models/olmo/modeling_olmo.py | 3 +++ .../models/olmo2/modeling_olmo2.py | 3 +++ src/transformers/utils/generic.py | 18 +++++++++++------- 14 files changed, 50 insertions(+), 7 deletions(-) diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index 24894b817bed..4d81cca0e3dd 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -756,6 +756,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/bitnet/modeling_bitnet.py b/src/transformers/models/bitnet/modeling_bitnet.py index dd6575adc643..94db9c9d48de 100644 --- a/src/transformers/models/bitnet/modeling_bitnet.py +++ b/src/transformers/models/bitnet/modeling_bitnet.py @@ -370,6 +370,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/cohere/modeling_cohere.py b/src/transformers/models/cohere/modeling_cohere.py index 444e026ef768..744c5a217161 100644 --- a/src/transformers/models/cohere/modeling_cohere.py +++ b/src/transformers/models/cohere/modeling_cohere.py @@ -412,6 +412,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/csm/modeling_csm.py b/src/transformers/models/csm/modeling_csm.py index 67c8365b5881..a2c713359bf8 100644 --- a/src/transformers/models/csm/modeling_csm.py +++ b/src/transformers/models/csm/modeling_csm.py @@ -773,6 +773,9 @@ def forward( [What are input IDs?](../glossary#input-ids) """ + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py index 096aa4bfdde0..38ea230e5980 100644 --- a/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py +++ b/src/transformers/models/deepseek_v3/modeling_deepseek_v3.py @@ -560,6 +560,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 4a44f081dac7..3b64097af3df 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -645,6 +645,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index c0c01b34a17d..0e39ddf243c5 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -1232,6 +1232,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index 0656ac4e827a..a7cf1865d873 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -390,6 +390,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index 7524006a293d..b31d1c2315c7 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -403,6 +403,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index f164ac5a2ddc..d55750d8faf0 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -375,6 +375,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index c421abf04c74..b4ee0603f174 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -379,6 +379,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/olmo/modeling_olmo.py b/src/transformers/models/olmo/modeling_olmo.py index 4225ba00b97f..731dbf43abbf 100644 --- a/src/transformers/models/olmo/modeling_olmo.py +++ b/src/transformers/models/olmo/modeling_olmo.py @@ -351,6 +351,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/models/olmo2/modeling_olmo2.py b/src/transformers/models/olmo2/modeling_olmo2.py index 11598c1ad872..9d8f573ed1f7 100644 --- a/src/transformers/models/olmo2/modeling_olmo2.py +++ b/src/transformers/models/olmo2/modeling_olmo2.py @@ -362,6 +362,9 @@ def forward( use_cache: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: + if (input_ids is None) ^ (inputs_embeds is not None): + raise ValueError("You must specify exactly one of input_ids or inputs_embeds") + if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index a5a8c67bc124..9dab343f65da 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -986,19 +986,23 @@ def output_hidden_and_attention(module, inp, out): if output_attentions: collected_attentions.append(out[1]) - for _, layer in self.named_modules(): - if isinstance(layer, GradientCheckpointingLayer): # TODO a bit slow to iterate over all layers no? - hooks.append(layer.register_forward_hook(output_hidden_and_attention)) + def capture_outputs(module, input, output): + if output_hidden_states: + collected_hidden_states.append(output[0]) + if output_attentions: + collected_attentions.append(output[1]) - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You must specify exactly one of input_ids or inputs_embeds.") + # Register hooks if needed + if output_attentions or output_hidden_states: + with torch.compiler.disable(): # Avoid issues with torch.compile + for _, layer in self.named_modules(): + if isinstance(layer, GradientCheckpointingLayer): + hooks.append(layer.register_forward_hook(capture_outputs)) if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") kwargs["use_cache"] = False # update it directly in kwargs - if past_key_values is not None and "Cache" not in past_key_values.__class__.__name__: - raise ValueError("The `past_key_values` should be either a `Cache` object or `None`.") outputs = func(self, *args, **kwargs) if output_attentions or output_hidden_states: From 19ef67ad7b145d235e4b7cda5a53a6f5d92f51de Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 15:55:32 +0200 Subject: [PATCH 26/51] oups --- src/transformers/utils/generic.py | 11 +++-------- 1 file changed, 3 insertions(+), 8 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 9dab343f65da..7e0d5e8326d3 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -994,14 +994,9 @@ def capture_outputs(module, input, output): # Register hooks if needed if output_attentions or output_hidden_states: - with torch.compiler.disable(): # Avoid issues with torch.compile - for _, layer in self.named_modules(): - if isinstance(layer, GradientCheckpointingLayer): - hooks.append(layer.register_forward_hook(capture_outputs)) - - if getattr(self, "gradient_checkpointing", False) and getattr(self, "training", False) and use_cache: - logger.warning("`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`.") - kwargs["use_cache"] = False # update it directly in kwargs + for _, layer in self.named_modules(): + if isinstance(layer, GradientCheckpointingLayer): + hooks.append(layer.register_forward_hook(capture_outputs)) outputs = func(self, *args, **kwargs) From 8fcc791590770b2cc098925c5bcb1a876d3b86dc Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:01:48 +0200 Subject: [PATCH 27/51] pass compile exigences --- src/transformers/utils/generic.py | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 7e0d5e8326d3..a931a7a061d1 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -49,6 +49,8 @@ import torch # noqa: F401 from ..modeling_layers import GradientCheckpointingLayer + from torch._dynamo import is_compiling, disable + class cached_property(property): @@ -943,6 +945,14 @@ def wrapper(self, *args, **kwargs): return wrapper +def register_hook_if_needed(layer, capture_outputs): + if is_compiling(): + pass + # TorchDynamo is tracing — wrap in disable context + else: + # Eager mode — no need to disable + return layer.register_forward_hook(capture_outputs) + def check_model_inputs(func): """ Decorator to check if the model inputs are valid before calling the function. @@ -996,13 +1006,14 @@ def capture_outputs(module, input, output): if output_attentions or output_hidden_states: for _, layer in self.named_modules(): if isinstance(layer, GradientCheckpointingLayer): - hooks.append(layer.register_forward_hook(capture_outputs)) + hooks.append(register_hook_if_needed(layer, capture_outputs)) outputs = func(self, *args, **kwargs) if output_attentions or output_hidden_states: for h in hooks: - h.remove() + if h is not None: + h.remove() if output_hidden_states: collected_hidden_states.append(outputs.last_hidden_state) From 39322ed14519627a9660206a6c7c808363e93b5c Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:12:26 +0200 Subject: [PATCH 28/51] fixes, and update mixtral --- .../models/llama/modeling_llama.py | 2 + .../models/mixtral/modular_mixtral.py | 93 ++----------------- 2 files changed, 9 insertions(+), 86 deletions(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index b4ee0603f174..b3fc2b9c8424 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -573,6 +573,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -731,6 +732,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/mixtral/modular_mixtral.py b/src/transformers/models/mixtral/modular_mixtral.py index 221aeb380a9d..afb221cc19dc 100644 --- a/src/transformers/models/mixtral/modular_mixtral.py +++ b/src/transformers/models/mixtral/modular_mixtral.py @@ -152,17 +152,6 @@ def forward(self, hidden_states): class MixtralSparseMoeBlock(nn.Module): - """ - This implementation is - strictly equivalent to standard MoE with full capacity (no - dropped tokens). It's faster since it formulates MoE operations - in terms of block-sparse operations to accommodate imbalanced - assignments of tokens to experts, whereas standard MoE either - (1) drop tokens at the cost of reduced performance or (2) set - capacity factor to number of experts and thus waste computation - and memory on padding. - """ - def __init__(self, config): super().__init__() self.hidden_dim = config.hidden_size @@ -293,18 +282,10 @@ def forward( # Fully Connected residual = hidden_states hidden_states = self.post_attention_layernorm(hidden_states) - hidden_states, router_logits = self.block_sparse_moe(hidden_states) + hidden_states = self.block_sparse_moe(hidden_states)[0] hidden_states = residual + hidden_states - outputs = (hidden_states,) - - if output_attentions: - outputs += (self_attn_weights,) - - if output_router_logits: - outputs += (router_logits,) - - return outputs + return outputs, self_attn_weights class MixtralRotaryEmbedding(MistralRotaryEmbedding): @@ -316,6 +297,10 @@ class MixtralPreTrainedModel(MistralPreTrainedModel): class MixtralModel(MistralModel): + _can_record_outputs = { + "roouter_logits": (MixtralBlockSparseTop2MLP, 1) + } + def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -324,31 +309,13 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_router_logits = ( - output_router_logits if output_router_logits is not None else self.config.output_router_logits - ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - if use_cache and past_key_values is None: past_key_values = DynamicCache() @@ -376,30 +343,9 @@ def forward( # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - all_router_logits = () if output_router_logits else None - for decoder_layer in self.layers: - if output_hidden_states: - all_hidden_states += (hidden_states,) - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **kwargs), - hidden_states, - causal_mask, - position_ids, - past_key_values, - output_attentions, - output_router_logits, - use_cache, - cache_position, - position_embeddings, - ) - else: + layer_outputs = decoder_layer( hidden_states, attention_mask=causal_mask, @@ -415,12 +361,6 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - all_self_attns += (layer_outputs[1],) - - if output_router_logits: - all_router_logits += (layer_outputs[-1],) - hidden_states = self.norm(hidden_states) # add hidden states from the last decoder layer @@ -430,9 +370,6 @@ def forward( return MoeModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=past_key_values, - hidden_states=all_hidden_states, - attentions=all_self_attns, - router_logits=all_router_logits, ) @@ -458,9 +395,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -487,16 +421,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_router_logits = ( - output_router_logits if output_router_logits is not None else self.config.output_router_logits - ) - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( input_ids=input_ids, @@ -505,9 +429,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - output_router_logits=output_router_logits, cache_position=cache_position, **kwargs, ) From 7d0ab1a12977ef9a2d5690a3781871c6e3eee3b4 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:25:17 +0200 Subject: [PATCH 29/51] more advanced api --- src/transformers/modeling_utils.py | 7 ++++- src/transformers/utils/generic.py | 48 ++++++++++++++++-------------- 2 files changed, 31 insertions(+), 24 deletions(-) diff --git a/src/transformers/modeling_utils.py b/src/transformers/modeling_utils.py index ae6f194a90b4..e4899cd891f9 100644 --- a/src/transformers/modeling_utils.py +++ b/src/transformers/modeling_utils.py @@ -136,7 +136,7 @@ is_torchdynamo_compiling, ) from .utils.quantization_config import BitsAndBytesConfig, QuantizationMethod - +from .modeling_layers import GradientCheckpointingLayer XLA_USE_BF16 = os.environ.get("XLA_USE_BF16", "0").upper() XLA_DOWNCAST_BF16 = os.environ.get("XLA_DOWNCAST_BF16", "0").upper() @@ -2001,6 +2001,7 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, PushToHubMixin, PeftAdapterMi # In practice, it means that they support attention interface functions, fully pass the kwargs # through all modules up to the Attention layer, can slice logits with Tensor, and have a default TP plan _supports_attention_backend = False + _can_record_outputs = None @property def dummy_inputs(self) -> dict[str, torch.Tensor]: @@ -2050,6 +2051,10 @@ def __init__(self, config: PretrainedConfig, *inputs, **kwargs): self._keep_in_fp32_modules = copy.copy(self.__class__._keep_in_fp32_modules) self._no_split_modules = self._no_split_modules or [] + self._can_record_outputs = { + "attentions": (GradientCheckpointingLayer, 1), + "hidden_states": (GradientCheckpointingLayer, 0), + } def post_init(self): """ diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index a931a7a061d1..41329d1666ad 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -996,34 +996,36 @@ def output_hidden_and_attention(module, inp, out): if output_attentions: collected_attentions.append(out[1]) - def capture_outputs(module, input, output): - if output_hidden_states: - collected_hidden_states.append(output[0]) - if output_attentions: - collected_attentions.append(output[1]) + def make_capture_fn(key, index): + def capture_fn(module, input, output): + collected[key].append(output[index]) + return capture_fn - # Register hooks if needed - if output_attentions or output_hidden_states: + if any(capture_flags.values()): for _, layer in self.named_modules(): - if isinstance(layer, GradientCheckpointingLayer): - hooks.append(register_hook_if_needed(layer, capture_outputs)) + for key, (cls, idx) in self._can_record_outputs.items(): + if capture_flags.get(key, False) and isinstance(layer, cls): + hook_fn = make_capture_fn(key, idx) + hooks.append(register_hook_if_needed(layer, hook_fn)) outputs = func(self, *args, **kwargs) - if output_attentions or output_hidden_states: - for h in hooks: - if h is not None: - h.remove() - - if output_hidden_states: - collected_hidden_states.append(outputs.last_hidden_state) - outputs["hidden_states"] = tuple(collected_hidden_states) - elif not return_dict: - outputs["hidden_states"] = None - if output_attentions: - outputs["attentions"] = tuple(collected_attentions) - elif not return_dict: - outputs["attentions"] = None + for h in hooks: + if h is not None: + h.remove() + + # Insert collected outputs + for key in collected: + # Optional: append a final model-level output like `last_hidden_state` for hidden_states + # if needed. You can customize this as per key semantics. + outputs[key] = tuple(collected[key]) if collected[key] else None + + # Fill missing keys with None if not using return_dict + if return_dict is False: + for key in recordable_keys: + if key not in outputs: + outputs[key] = None + if return_dict is False: outputs = outputs.to_tuple() return outputs From 2de1fcd039971f49bdd5bfd594f2adb3a41ca9c4 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:31:23 +0200 Subject: [PATCH 30/51] update --- src/transformers/modeling_utils.py | 5 ++- .../models/jetmoe/modeling_jetmoe.py | 1 + .../models/mixtral/modular_mixtral.py | 37 ++++++++----------- .../models/nemotron/modeling_nemotron.py | 2 + .../models/persimmon/modeling_persimmon.py | 2 + .../models/phimoe/modeling_phimoe.py | 1 + .../models/qwen2_moe/modeling_qwen2_moe.py | 2 + .../models/stablelm/modeling_stablelm.py | 2 + src/transformers/utils/generic.py | 8 ++-- 9 files changed, 32 insertions(+), 28 deletions(-) diff --git a/src/transformers/modeling_utils.py b/src/transformers/modeling_utils.py index e4899cd891f9..285a2fe0c70c 100644 --- a/src/transformers/modeling_utils.py +++ b/src/transformers/modeling_utils.py @@ -73,6 +73,7 @@ verify_tp_plan, ) from .loss.loss_utils import LOSS_MAPPING +from .modeling_layers import GradientCheckpointingLayer from .pytorch_utils import ( # noqa: F401 Conv1D, apply_chunking_to_forward, @@ -136,7 +137,7 @@ is_torchdynamo_compiling, ) from .utils.quantization_config import BitsAndBytesConfig, QuantizationMethod -from .modeling_layers import GradientCheckpointingLayer + XLA_USE_BF16 = os.environ.get("XLA_USE_BF16", "0").upper() XLA_DOWNCAST_BF16 = os.environ.get("XLA_DOWNCAST_BF16", "0").upper() @@ -2052,8 +2053,8 @@ def __init__(self, config: PretrainedConfig, *inputs, **kwargs): self._no_split_modules = self._no_split_modules or [] self._can_record_outputs = { - "attentions": (GradientCheckpointingLayer, 1), "hidden_states": (GradientCheckpointingLayer, 0), + "attentions": (GradientCheckpointingLayer, 1), } def post_init(self): diff --git a/src/transformers/models/jetmoe/modeling_jetmoe.py b/src/transformers/models/jetmoe/modeling_jetmoe.py index ab843099c54b..5753c616553c 100644 --- a/src/transformers/models/jetmoe/modeling_jetmoe.py +++ b/src/transformers/models/jetmoe/modeling_jetmoe.py @@ -1311,6 +1311,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/mixtral/modular_mixtral.py b/src/transformers/models/mixtral/modular_mixtral.py index afb221cc19dc..3a9e888fecb9 100644 --- a/src/transformers/models/mixtral/modular_mixtral.py +++ b/src/transformers/models/mixtral/modular_mixtral.py @@ -297,9 +297,7 @@ class MixtralPreTrainedModel(MistralPreTrainedModel): class MixtralModel(MistralModel): - _can_record_outputs = { - "roouter_logits": (MixtralBlockSparseTop2MLP, 1) - } + _can_record_outputs = {"roouter_logits": (MixtralBlockSparseTop2MLP, 1)} def forward( self, @@ -312,7 +310,6 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") @@ -344,22 +341,20 @@ def forward( # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) for decoder_layer in self.layers: + layer_outputs = decoder_layer( + hidden_states, + attention_mask=causal_mask, + position_ids=position_ids, + past_key_value=past_key_values, + output_attentions=output_attentions, + output_router_logits=output_router_logits, + use_cache=use_cache, + cache_position=cache_position, + position_embeddings=position_embeddings, + **kwargs, + ) - - layer_outputs = decoder_layer( - hidden_states, - attention_mask=causal_mask, - position_ids=position_ids, - past_key_value=past_key_values, - output_attentions=output_attentions, - output_router_logits=output_router_logits, - use_cache=use_cache, - cache_position=cache_position, - position_embeddings=position_embeddings, - **kwargs, - ) - - hidden_states = layer_outputs[0] + hidden_states = layer_outputs[0] hidden_states = self.norm(hidden_states) @@ -441,9 +436,9 @@ def forward( loss = None if labels is not None: loss = self.loss_function(logits, labels, self.vocab_size, **kwargs) - +"" aux_loss = None - if output_router_logits: + if outputs.router_logits is not None: aux_loss = load_balancing_loss_func( outputs.router_logits, self.num_experts, diff --git a/src/transformers/models/nemotron/modeling_nemotron.py b/src/transformers/models/nemotron/modeling_nemotron.py index 6b5cf370c9da..e1c1c82e0b47 100644 --- a/src/transformers/models/nemotron/modeling_nemotron.py +++ b/src/transformers/models/nemotron/modeling_nemotron.py @@ -1015,6 +1015,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -1175,6 +1176,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/persimmon/modeling_persimmon.py b/src/transformers/models/persimmon/modeling_persimmon.py index ce142c8d6d22..3b29d8e4e753 100644 --- a/src/transformers/models/persimmon/modeling_persimmon.py +++ b/src/transformers/models/persimmon/modeling_persimmon.py @@ -856,6 +856,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -951,6 +952,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/phimoe/modeling_phimoe.py b/src/transformers/models/phimoe/modeling_phimoe.py index 2a9240783399..f4d709343307 100644 --- a/src/transformers/models/phimoe/modeling_phimoe.py +++ b/src/transformers/models/phimoe/modeling_phimoe.py @@ -1443,6 +1443,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py index 9882ca447d7e..00db18aaf37b 100644 --- a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py +++ b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py @@ -1246,6 +1246,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -1341,6 +1342,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/stablelm/modeling_stablelm.py b/src/transformers/models/stablelm/modeling_stablelm.py index df4e41bcd210..a5399c90968a 100755 --- a/src/transformers/models/stablelm/modeling_stablelm.py +++ b/src/transformers/models/stablelm/modeling_stablelm.py @@ -1083,6 +1083,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -1178,6 +1179,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 41329d1666ad..b83ae118b855 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -47,10 +47,7 @@ if is_torch_available(): # required for @can_return_tuple decorator to work with torchdynamo import torch # noqa: F401 - - from ..modeling_layers import GradientCheckpointingLayer - from torch._dynamo import is_compiling, disable - + from torch._dynamo import is_compiling class cached_property(property): @@ -953,6 +950,7 @@ def register_hook_if_needed(layer, capture_outputs): # Eager mode — no need to disable return layer.register_forward_hook(capture_outputs) + def check_model_inputs(func): """ Decorator to check if the model inputs are valid before calling the function. @@ -999,6 +997,7 @@ def output_hidden_and_attention(module, inp, out): def make_capture_fn(key, index): def capture_fn(module, input, output): collected[key].append(output[index]) + return capture_fn if any(capture_flags.values()): @@ -1008,7 +1007,6 @@ def capture_fn(module, input, output): hook_fn = make_capture_fn(key, idx) hooks.append(register_hook_if_needed(layer, hook_fn)) - outputs = func(self, *args, **kwargs) for h in hooks: if h is not None: From 531dc0c2865295520cc5c1c9a5ba015b7d4c2305 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:37:16 +0200 Subject: [PATCH 31/51] fix return dict not even present --- .../models/diffllama/modeling_diffllama.py | 2 + .../models/gemma/modeling_gemma.py | 2 + .../models/gemma2/modeling_gemma2.py | 2 + src/transformers/models/glm/modeling_glm.py | 2 + src/transformers/models/glm4/modeling_glm4.py | 2 + .../models/helium/modeling_helium.py | 2 + .../models/mistral/modeling_mistral.py | 2 + .../models/mixtral/modeling_mixtral.py | 124 +++--------------- .../models/mixtral/modular_mixtral.py | 2 +- src/transformers/models/phi/modeling_phi.py | 2 + src/transformers/models/phi3/modeling_phi3.py | 2 + .../models/qwen2/modeling_qwen2.py | 2 + .../models/qwen3/modeling_qwen3.py | 2 + .../models/qwen3_moe/modeling_qwen3_moe.py | 101 +++++--------- .../models/starcoder2/modeling_starcoder2.py | 2 + 15 files changed, 82 insertions(+), 169 deletions(-) diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 3b64097af3df..25732f685943 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -836,6 +836,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -993,6 +994,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py index 9d5c0c129d56..4c6e1f848602 100644 --- a/src/transformers/models/gemma/modeling_gemma.py +++ b/src/transformers/models/gemma/modeling_gemma.py @@ -606,6 +606,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -700,6 +701,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/gemma2/modeling_gemma2.py b/src/transformers/models/gemma2/modeling_gemma2.py index 9cda62d400e1..bda7b81b3d43 100644 --- a/src/transformers/models/gemma2/modeling_gemma2.py +++ b/src/transformers/models/gemma2/modeling_gemma2.py @@ -671,6 +671,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -765,6 +766,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/glm/modeling_glm.py b/src/transformers/models/glm/modeling_glm.py index a7cf1865d873..6fd23ec62a8f 100644 --- a/src/transformers/models/glm/modeling_glm.py +++ b/src/transformers/models/glm/modeling_glm.py @@ -581,6 +581,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -675,6 +676,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/glm4/modeling_glm4.py b/src/transformers/models/glm4/modeling_glm4.py index b31d1c2315c7..ebfc041601bc 100644 --- a/src/transformers/models/glm4/modeling_glm4.py +++ b/src/transformers/models/glm4/modeling_glm4.py @@ -594,6 +594,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -688,6 +689,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/helium/modeling_helium.py b/src/transformers/models/helium/modeling_helium.py index d55750d8faf0..054ec8d67eae 100644 --- a/src/transformers/models/helium/modeling_helium.py +++ b/src/transformers/models/helium/modeling_helium.py @@ -566,6 +566,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -660,6 +661,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py index c6df5f22af3c..58e26c2943e7 100644 --- a/src/transformers/models/mistral/modeling_mistral.py +++ b/src/transformers/models/mistral/modeling_mistral.py @@ -575,6 +575,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -652,6 +653,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/mixtral/modeling_mixtral.py b/src/transformers/models/mixtral/modeling_mixtral.py index 4884de54d076..a4d2d95c792f 100644 --- a/src/transformers/models/mixtral/modeling_mixtral.py +++ b/src/transformers/models/mixtral/modeling_mixtral.py @@ -74,17 +74,6 @@ def forward(self, hidden_states): class MixtralSparseMoeBlock(nn.Module): - """ - This implementation is - strictly equivalent to standard MoE with full capacity (no - dropped tokens). It's faster since it formulates MoE operations - in terms of block-sparse operations to accommodate imbalanced - assignments of tokens to experts, whereas standard MoE either - (1) drop tokens at the cost of reduced performance or (2) set - capacity factor to number of experts and thus waste computation - and memory on padding. - """ - def __init__(self, config): super().__init__() self.hidden_dim = config.hidden_size @@ -362,18 +351,10 @@ def forward( # Fully Connected residual = hidden_states hidden_states = self.post_attention_layernorm(hidden_states) - hidden_states, router_logits = self.block_sparse_moe(hidden_states) + hidden_states = self.block_sparse_moe(hidden_states)[0] hidden_states = residual + hidden_states - outputs = (hidden_states,) - - if output_attentions: - outputs += (self_attn_weights,) - - if output_router_logits: - outputs += (router_logits,) - - return outputs + return outputs, self_attn_weights class MixtralRotaryEmbedding(nn.Module): @@ -441,6 +422,8 @@ def _init_weights(self, module): @auto_docstring class MixtralModel(MixtralPreTrainedModel): + _can_record_outputs = {"roouter_logits": (MixtralBlockSparseTop2MLP, 1)} + def __init__(self, config: MixtralConfig): super().__init__(config) self.padding_idx = config.pad_token_id @@ -473,31 +456,12 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_router_logits = ( - output_router_logits if output_router_logits is not None else self.config.output_router_logits - ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - if use_cache and past_key_values is None: past_key_values = DynamicCache() @@ -525,50 +489,21 @@ def forward( # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - all_router_logits = () if output_router_logits else None - for decoder_layer in self.layers: - if output_hidden_states: - all_hidden_states += (hidden_states,) - - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **kwargs), - hidden_states, - causal_mask, - position_ids, - past_key_values, - output_attentions, - output_router_logits, - use_cache, - cache_position, - position_embeddings, - ) - else: - layer_outputs = decoder_layer( - hidden_states, - attention_mask=causal_mask, - position_ids=position_ids, - past_key_value=past_key_values, - output_attentions=output_attentions, - output_router_logits=output_router_logits, - use_cache=use_cache, - cache_position=cache_position, - position_embeddings=position_embeddings, - **kwargs, - ) - - hidden_states = layer_outputs[0] - - if output_attentions: - all_self_attns += (layer_outputs[1],) - - if output_router_logits: - all_router_logits += (layer_outputs[-1],) + layer_outputs = decoder_layer( + hidden_states, + attention_mask=causal_mask, + position_ids=position_ids, + past_key_value=past_key_values, + output_attentions=output_attentions, + output_router_logits=output_router_logits, + use_cache=use_cache, + cache_position=cache_position, + position_embeddings=position_embeddings, + **kwargs, + ) + + hidden_states = layer_outputs[0] hidden_states = self.norm(hidden_states) @@ -579,9 +514,6 @@ def forward( return MoeModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=past_key_values, - hidden_states=all_hidden_states, - attentions=all_self_attns, - router_logits=all_router_logits, ) @@ -717,9 +649,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -746,16 +675,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_router_logits = ( - output_router_logits if output_router_logits is not None else self.config.output_router_logits - ) - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( input_ids=input_ids, @@ -764,9 +683,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - output_router_logits=output_router_logits, cache_position=cache_position, **kwargs, ) @@ -781,7 +697,7 @@ def forward( loss = self.loss_function(logits, labels, self.vocab_size, **kwargs) aux_loss = None - if output_router_logits: + if outputs.router_logits is not None: aux_loss = load_balancing_loss_func( outputs.router_logits, self.num_experts, @@ -845,6 +761,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -939,6 +856,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/mixtral/modular_mixtral.py b/src/transformers/models/mixtral/modular_mixtral.py index 3a9e888fecb9..d1a3aadc9279 100644 --- a/src/transformers/models/mixtral/modular_mixtral.py +++ b/src/transformers/models/mixtral/modular_mixtral.py @@ -436,7 +436,7 @@ def forward( loss = None if labels is not None: loss = self.loss_function(logits, labels, self.vocab_size, **kwargs) -"" + aux_loss = None if outputs.router_logits is not None: aux_loss = load_balancing_loss_func( diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index 2a3a005d8050..7b63ebbc6dad 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -597,6 +597,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -691,6 +692,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index 47833949c96b..b708bcb999b5 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -680,6 +680,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -774,6 +775,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py index 058e7eb46f4b..d81faaf7a612 100644 --- a/src/transformers/models/qwen2/modeling_qwen2.py +++ b/src/transformers/models/qwen2/modeling_qwen2.py @@ -595,6 +595,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -689,6 +690,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 517cf1c5cc3d..27190d6ba524 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -624,6 +624,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -718,6 +719,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py index 6a084d2022cd..d5674ba7d953 100644 --- a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py +++ b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py @@ -382,6 +382,24 @@ def forward( return outputs +class Qwen3MoeBlockSparseTop2MLP(nn.Module): + def __init__(self, config: Qwen3MoeConfig): + super().__init__() + self.ffn_dim = config.intermediate_size + self.hidden_dim = config.hidden_size + + self.w1 = nn.Linear(self.hidden_dim, self.ffn_dim, bias=False) + self.w2 = nn.Linear(self.ffn_dim, self.hidden_dim, bias=False) + self.w3 = nn.Linear(self.hidden_dim, self.ffn_dim, bias=False) + + self.act_fn = ACT2FN[config.hidden_act] + + def forward(self, hidden_states): + current_hidden_states = self.act_fn(self.w1(hidden_states)) * self.w3(hidden_states) + current_hidden_states = self.w2(current_hidden_states) + return current_hidden_states + + class Qwen3MoeRotaryEmbedding(nn.Module): def __init__(self, config: Qwen3MoeConfig, device=None): super().__init__() @@ -447,6 +465,8 @@ def _init_weights(self, module): @auto_docstring class Qwen3MoeModel(Qwen3MoePreTrainedModel): + _can_record_outputs = {"roouter_logits": (Qwen3MoeBlockSparseTop2MLP, 1)} + def __init__(self, config: Qwen3MoeConfig): super().__init__(config) self.padding_idx = config.pad_token_id @@ -479,31 +499,12 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_router_logits = ( - output_router_logits if output_router_logits is not None else self.config.output_router_logits - ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - if use_cache and past_key_values is None: past_key_values = DynamicCache() @@ -531,50 +532,21 @@ def forward( # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - all_router_logits = () if output_router_logits else None - for decoder_layer in self.layers: - if output_hidden_states: - all_hidden_states += (hidden_states,) - - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - partial(decoder_layer.__call__, **kwargs), - hidden_states, - causal_mask, - position_ids, - past_key_values, - output_attentions, - output_router_logits, - use_cache, - cache_position, - position_embeddings, - ) - else: - layer_outputs = decoder_layer( - hidden_states, - attention_mask=causal_mask, - position_ids=position_ids, - past_key_value=past_key_values, - output_attentions=output_attentions, - output_router_logits=output_router_logits, - use_cache=use_cache, - cache_position=cache_position, - position_embeddings=position_embeddings, - **kwargs, - ) - - hidden_states = layer_outputs[0] - - if output_attentions: - all_self_attns += (layer_outputs[1],) - - if output_router_logits: - all_router_logits += (layer_outputs[-1],) + layer_outputs = decoder_layer( + hidden_states, + attention_mask=causal_mask, + position_ids=position_ids, + past_key_value=past_key_values, + output_attentions=output_attentions, + output_router_logits=output_router_logits, + use_cache=use_cache, + cache_position=cache_position, + position_embeddings=position_embeddings, + **kwargs, + ) + + hidden_states = layer_outputs[0] hidden_states = self.norm(hidden_states) @@ -585,9 +557,6 @@ def forward( return MoeModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=past_key_values, - hidden_states=all_hidden_states, - attentions=all_self_attns, - router_logits=all_router_logits, ) @@ -851,6 +820,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -945,6 +915,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py index a662bc78eb8f..58f618b3d20f 100644 --- a/src/transformers/models/starcoder2/modeling_starcoder2.py +++ b/src/transformers/models/starcoder2/modeling_starcoder2.py @@ -585,6 +585,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -679,6 +680,7 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): From 168db36cd8839dfbfb2c03c837fe6442c357e6e5 Mon Sep 17 00:00:00 2001 From: Arthur Zucker Date: Wed, 28 May 2025 18:47:39 +0200 Subject: [PATCH 32/51] models that we should update --- .../models/albert/modeling_albert.py | 21 ++++------- .../models/align/modeling_align.py | 9 ++--- .../models/altclip/modeling_altclip.py | 21 ++++------- src/transformers/models/aria/modular_aria.py | 6 ++-- .../modeling_audio_spectrogram_transformer.py | 6 ++-- .../models/autoformer/modeling_autoformer.py | 9 ++--- .../models/aya_vision/modular_aya_vision.py | 3 +- .../models/bamba/modular_bamba.py | 6 ++-- src/transformers/models/bark/modeling_bark.py | 6 ++-- src/transformers/models/bart/modeling_bart.py | 21 ++++------- src/transformers/models/beit/modeling_beit.py | 12 +++---- src/transformers/models/bert/modeling_bert.py | 27 +++++--------- .../modeling_bert_generation.py | 6 ++-- .../models/big_bird/modeling_big_bird.py | 24 +++++-------- .../modeling_bigbird_pegasus.py | 21 ++++------- .../models/biogpt/modular_biogpt.py | 12 +++---- .../models/blenderbot/modeling_blenderbot.py | 9 ++--- .../modeling_blenderbot_small.py | 9 ++--- src/transformers/models/blip/modeling_blip.py | 18 ++++------ .../models/blip/modeling_blip_text.py | 6 ++-- .../models/blip_2/modeling_blip_2.py | 33 ++++++----------- .../models/bloom/modeling_bloom.py | 15 +++----- .../bridgetower/modeling_bridgetower.py | 12 +++---- src/transformers/models/bros/modeling_bros.py | 12 +++---- .../models/camembert/modeling_camembert.py | 21 ++++------- .../models/canine/modeling_canine.py | 15 +++----- .../models/chameleon/modeling_chameleon.py | 3 +- .../chinese_clip/modeling_chinese_clip.py | 21 ++++------- src/transformers/models/clap/modeling_clap.py | 21 ++++------- src/transformers/models/clip/modeling_clip.py | 33 ++++++----------- .../models/clipseg/modeling_clipseg.py | 30 ++++++---------- src/transformers/models/clvp/modeling_clvp.py | 12 +++---- .../models/codegen/modeling_codegen.py | 6 ++-- .../models/cohere/modular_cohere.py | 3 +- .../models/cohere2/modular_cohere2.py | 3 +- .../models/colpali/modeling_colpali.py | 3 +- .../modeling_conditional_detr.py | 3 +- .../models/convbert/modeling_convbert.py | 18 ++++------ .../models/cpmant/modeling_cpmant.py | 3 +- src/transformers/models/csm/modular_csm.py | 9 ++--- src/transformers/models/ctrl/modeling_ctrl.py | 9 ++--- .../models/dab_detr/modeling_dab_detr.py | 9 ++--- .../data2vec/modeling_data2vec_audio.py | 15 +++----- .../models/data2vec/modeling_data2vec_text.py | 21 ++++------- .../data2vec/modeling_data2vec_vision.py | 9 ++--- src/transformers/models/dbrx/modeling_dbrx.py | 6 ++-- .../models/deberta/modeling_deberta.py | 15 +++----- .../models/deberta_v2/modeling_deberta_v2.py | 18 ++++------ .../modeling_decision_transformer.py | 3 +- .../modeling_deformable_detr.py | 3 +- src/transformers/models/deit/modeling_deit.py | 12 +++---- .../depth_anything/modeling_depth_anything.py | 3 +- .../models/depth_pro/modeling_depth_pro.py | 6 ++-- src/transformers/models/detr/modeling_detr.py | 3 +- .../models/diffllama/modeling_diffllama.py | 9 ++--- .../models/dinat/modeling_dinat.py | 6 ++-- .../models/dinov2/modeling_dinov2.py | 6 ++-- .../modular_dinov2_with_registers.py | 3 +- .../models/distilbert/modeling_distilbert.py | 18 ++++------ .../models/donut/modeling_donut_swin.py | 6 ++-- src/transformers/models/dpr/modeling_dpr.py | 9 ++--- src/transformers/models/dpt/modeling_dpt.py | 9 ++--- .../models/electra/modeling_electra.py | 24 +++++-------- src/transformers/models/emu3/modular_emu3.py | 6 ++-- .../modeling_encoder_decoder.py | 3 +- .../models/ernie/modeling_ernie.py | 27 +++++--------- src/transformers/models/esm/modeling_esm.py | 12 +++---- .../models/falcon/modeling_falcon.py | 15 +++----- .../models/falcon_h1/modular_falcon_h1.py | 6 ++-- .../models/flaubert/modeling_flaubert.py | 21 ++++------- .../models/flava/modeling_flava.py | 15 +++----- src/transformers/models/fsmt/modeling_fsmt.py | 6 ++-- .../models/funnel/modeling_funnel.py | 24 +++++-------- .../models/gemma/modular_gemma.py | 3 +- .../models/gemma2/modular_gemma2.py | 6 ++-- .../models/gemma3/modular_gemma3.py | 9 ++--- .../models/got_ocr2/modular_got_ocr2.py | 6 ++-- src/transformers/models/gpt2/modeling_gpt2.py | 18 ++++------ .../models/gpt_neo/modeling_gpt_neo.py | 15 +++----- .../models/gpt_neox/modular_gpt_neox.py | 15 +++----- src/transformers/models/gptj/modeling_gptj.py | 12 +++---- .../models/granite/modular_granite.py | 6 ++-- .../modular_granitemoehybrid.py | 3 +- .../models/groupvit/modeling_groupvit.py | 21 ++++------- .../models/hubert/modular_hubert.py | 3 +- .../models/ijepa/modular_ijepa.py | 3 +- .../models/informer/modular_informer.py | 3 +- .../modular_instructblipvideo.py | 6 ++-- .../models/internvl/modular_internvl.py | 3 +- .../models/janus/modular_janus.py | 6 ++-- .../models/layoutlm/modeling_layoutlm.py | 15 +++----- .../models/layoutlmv2/modeling_layoutlmv2.py | 12 +++---- .../models/layoutlmv3/modeling_layoutlmv3.py | 12 +++---- src/transformers/models/led/modeling_led.py | 12 +++---- src/transformers/models/lilt/modeling_lilt.py | 12 +++---- .../models/llama/modeling_llama.py | 9 ++--- .../modular_llava_next_video.py | 6 ++-- .../modular_llava_onevision.py | 6 ++-- .../models/longformer/modeling_longformer.py | 18 ++++------ .../models/marian/modeling_marian.py | 15 +++----- .../models/mistral/modular_mistral.py | 6 ++-- .../models/mistral3/modular_mistral3.py | 6 ++-- src/transformers/models/mlcd/modular_mlcd.py | 9 ++--- .../models/mllama/modeling_mllama.py | 18 ++++------ .../models/modernbert/modular_modernbert.py | 15 +++----- .../models/moonshine/modular_moonshine.py | 12 +++---- src/transformers/models/mpt/modeling_mpt.py | 15 +++----- src/transformers/models/mt5/modeling_mt5.py | 18 ++++------ .../models/musicgen/modeling_musicgen.py | 12 +++---- src/transformers/models/mvp/modeling_mvp.py | 21 ++++------- .../models/nemotron/modeling_nemotron.py | 15 +++----- .../models/nllb_moe/modeling_nllb_moe.py | 12 +++---- .../models/olmoe/modeling_olmoe.py | 6 ++-- .../omdet_turbo/modeling_omdet_turbo.py | 3 +- .../models/openai/modeling_openai.py | 12 +++---- src/transformers/models/opt/modeling_opt.py | 15 +++----- .../models/owlv2/modeling_owlv2.py | 36 +++++++------------ .../models/owlvit/modeling_owlvit.py | 36 +++++++------------ .../models/paligemma/modeling_paligemma.py | 6 ++-- .../models/pegasus/modeling_pegasus.py | 9 ++--- .../models/pegasus_x/modeling_pegasus_x.py | 6 ++-- .../models/perceiver/modeling_perceiver.py | 24 +++++-------- .../models/persimmon/modeling_persimmon.py | 12 +++---- src/transformers/models/phi/modular_phi.py | 3 +- src/transformers/models/phi3/modeling_phi3.py | 9 ++--- .../modular_phi4_multimodal.py | 9 ++--- .../models/phimoe/modeling_phimoe.py | 9 ++--- .../models/pix2struct/modeling_pix2struct.py | 9 ++--- .../models/plbart/modular_plbart.py | 6 ++-- .../models/pop2piano/modeling_pop2piano.py | 3 +- .../modular_prompt_depth_anything.py | 3 +- .../models/prophetnet/modeling_prophetnet.py | 15 +++----- .../models/pvt_v2/modeling_pvt_v2.py | 9 ++--- .../models/qwen2/modular_qwen2.py | 3 +- .../qwen2_5_omni/modular_qwen2_5_omni.py | 6 ++-- .../models/qwen2_5_vl/modular_qwen2_5_vl.py | 3 +- .../qwen2_audio/modeling_qwen2_audio.py | 3 +- .../models/qwen2_moe/modeling_qwen2_moe.py | 15 +++----- .../models/qwen2_vl/modeling_qwen2_vl.py | 6 ++-- .../models/qwen3/modeling_qwen3.py | 12 +++---- .../models/qwen3_moe/modular_qwen3_moe.py | 3 +- src/transformers/models/rag/modeling_rag.py | 9 ++--- .../models/rembert/modeling_rembert.py | 21 ++++------- .../models/roberta/modeling_roberta.py | 21 ++++------- .../modeling_roberta_prelayernorm.py | 21 ++++------- .../models/roc_bert/modeling_roc_bert.py | 24 +++++-------- .../models/roformer/modeling_roformer.py | 21 ++++------- src/transformers/models/rwkv/modeling_rwkv.py | 6 ++-- src/transformers/models/sam/modeling_sam.py | 12 +++---- .../models/sam_hq/modular_sam_hq.py | 9 ++--- .../seamless_m4t/modeling_seamless_m4t.py | 30 ++++++---------- .../modeling_seamless_m4t_v2.py | 33 ++++++----------- .../models/segformer/modeling_segformer.py | 9 ++--- .../models/seggpt/modeling_seggpt.py | 6 ++-- src/transformers/models/sew/modular_sew.py | 3 +- .../models/sew_d/modeling_sew_d.py | 9 ++--- .../shieldgemma2/modeling_shieldgemma2.py | 3 +- .../models/siglip/modeling_siglip.py | 27 +++++--------- .../models/siglip2/modular_siglip2.py | 15 +++----- .../models/smolvlm/modular_smolvlm.py | 3 +- .../modeling_speech_encoder_decoder.py | 3 +- .../speech_to_text/modeling_speech_to_text.py | 6 ++-- .../models/speecht5/modeling_speecht5.py | 36 +++++++------------ .../models/splinter/modeling_splinter.py | 9 ++--- .../squeezebert/modeling_squeezebert.py | 18 ++++------ .../models/stablelm/modeling_stablelm.py | 12 +++---- .../models/starcoder2/modular_starcoder2.py | 3 +- .../models/superglue/modeling_superglue.py | 3 +- src/transformers/models/swin/modeling_swin.py | 9 ++--- .../models/swin2sr/modeling_swin2sr.py | 6 ++-- .../models/swinv2/modeling_swinv2.py | 12 +++---- .../modeling_switch_transformers.py | 9 ++--- src/transformers/models/t5/modeling_t5.py | 18 ++++------ .../modeling_table_transformer.py | 3 +- .../models/tapas/modeling_tapas.py | 12 +++---- .../modeling_time_series_transformer.py | 9 ++--- .../models/timesfm/modular_timesfm.py | 3 +- .../timesformer/modeling_timesformer.py | 6 ++-- .../timm_backbone/modeling_timm_backbone.py | 3 +- .../models/trocr/modeling_trocr.py | 3 +- src/transformers/models/tvp/modeling_tvp.py | 9 ++--- src/transformers/models/udop/modeling_udop.py | 9 ++--- src/transformers/models/umt5/modeling_umt5.py | 18 ++++------ .../models/unispeech/modular_unispeech.py | 6 ++-- .../unispeech_sat/modular_unispeech_sat.py | 6 ++-- .../models/upernet/modeling_upernet.py | 3 +- .../video_llava/modeling_video_llava.py | 6 ++-- .../models/videomae/modeling_videomae.py | 9 ++--- src/transformers/models/vilt/modeling_vilt.py | 18 ++++------ .../models/vipllava/modular_vipllava.py | 6 ++-- .../modeling_vision_encoder_decoder.py | 3 +- .../modeling_vision_text_dual_encoder.py | 3 +- .../visual_bert/modeling_visual_bert.py | 18 ++++------ src/transformers/models/vit/modeling_vit.py | 9 ++--- .../models/vit_mae/modeling_vit_mae.py | 6 ++-- .../models/vit_msn/modeling_vit_msn.py | 6 ++-- .../models/vitdet/modeling_vitdet.py | 3 +- .../models/vitmatte/modeling_vitmatte.py | 3 +- .../models/vitpose/modeling_vitpose.py | 3 +- .../modeling_vitpose_backbone.py | 3 +- src/transformers/models/vits/modeling_vits.py | 9 ++--- .../models/vivit/modeling_vivit.py | 6 ++-- .../models/wav2vec2/modeling_wav2vec2.py | 21 ++++------- .../wav2vec2_bert/modular_wav2vec2_bert.py | 15 +++----- .../modeling_wav2vec2_conformer.py | 18 ++++------ .../models/wavlm/modeling_wavlm.py | 15 +++----- .../models/whisper/modeling_whisper.py | 12 +++---- .../models/x_clip/modeling_x_clip.py | 30 ++++++---------- src/transformers/models/xglm/modeling_xglm.py | 6 ++-- src/transformers/models/xlm/modeling_xlm.py | 21 ++++------- .../xlm_roberta/modeling_xlm_roberta.py | 21 ++++------- .../xlm_roberta_xl/modeling_xlm_roberta_xl.py | 21 ++++------- .../models/xlnet/modeling_xlnet.py | 21 ++++------- src/transformers/models/xmod/modeling_xmod.py | 21 ++++------- .../models/yolos/modeling_yolos.py | 3 +- src/transformers/models/yoso/modeling_yoso.py | 18 ++++------ .../models/zamba/modeling_zamba.py | 9 ++--- .../models/zamba2/modular_zamba2.py | 3 +- .../models/zoedepth/modeling_zoedepth.py | 3 +- 219 files changed, 825 insertions(+), 1650 deletions(-) diff --git a/src/transformers/models/albert/modeling_albert.py b/src/transformers/models/albert/modeling_albert.py index 1162c16f61cf..fdd3a9bce1af 100755 --- a/src/transformers/models/albert/modeling_albert.py +++ b/src/transformers/models/albert/modeling_albert.py @@ -663,8 +663,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPooling, tuple]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -781,8 +780,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, sentence_order_label: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[AlbertForPreTrainingOutput, tuple]: r""" @@ -926,8 +924,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MaskedLMOutput, tuple]: r""" @@ -1028,8 +1025,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple]: r""" @@ -1120,8 +1116,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[TokenClassifierOutput, tuple]: r""" @@ -1187,8 +1182,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[AlbertForPreTrainingOutput, tuple]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1264,8 +1258,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[AlbertForPreTrainingOutput, tuple]: r""" diff --git a/src/transformers/models/align/modeling_align.py b/src/transformers/models/align/modeling_align.py index 952fe0bdc9e4..f8fe9f658559 100644 --- a/src/transformers/models/align/modeling_align.py +++ b/src/transformers/models/align/modeling_align.py @@ -1088,8 +1088,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -1307,8 +1306,7 @@ def get_text_features( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1407,8 +1405,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, AlignOutput]: r""" diff --git a/src/transformers/models/altclip/modeling_altclip.py b/src/transformers/models/altclip/modeling_altclip.py index 3e917940809c..ff8d16e8c777 100755 --- a/src/transformers/models/altclip/modeling_altclip.py +++ b/src/transformers/models/altclip/modeling_altclip.py @@ -802,8 +802,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1031,8 +1030,7 @@ def __init__(self, config: AltCLIPVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = False, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -1087,8 +1085,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -1185,8 +1182,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1414,8 +1410,7 @@ def get_text_features( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, token_type_ids=None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1458,8 +1453,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> torch.FloatTensor: @@ -1511,8 +1505,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, token_type_ids: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, AltCLIPOutput]: diff --git a/src/transformers/models/aria/modular_aria.py b/src/transformers/models/aria/modular_aria.py index b5c18a40b715..0319e1922b2e 100644 --- a/src/transformers/models/aria/modular_aria.py +++ b/src/transformers/models/aria/modular_aria.py @@ -1425,8 +1425,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1522,8 +1521,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, logits_to_keep: Union[int, torch.Tensor] = 0, cache_position: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py index d3ccf24153b7..9aa49d1eb2a5 100644 --- a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py +++ b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py @@ -436,8 +436,7 @@ def forward( self, input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -526,8 +525,7 @@ def forward( input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/autoformer/modeling_autoformer.py b/src/transformers/models/autoformer/modeling_autoformer.py index 6db63b4945f7..be15d09293e6 100644 --- a/src/transformers/models/autoformer/modeling_autoformer.py +++ b/src/transformers/models/autoformer/modeling_autoformer.py @@ -969,8 +969,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1113,8 +1112,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, AutoFormerDecoderOutput]: r""" @@ -1993,8 +1991,7 @@ def generate( past_observed_mask: Optional[torch.Tensor] = None, static_categorical_features: Optional[torch.Tensor] = None, static_real_features: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> SampleTSPredictionOutput: r""" Greedily generate sequences of sample predictions from a model with a probability distribution head. diff --git a/src/transformers/models/aya_vision/modular_aya_vision.py b/src/transformers/models/aya_vision/modular_aya_vision.py index ad5c1e58d438..727f917ce551 100644 --- a/src/transformers/models/aya_vision/modular_aya_vision.py +++ b/src/transformers/models/aya_vision/modular_aya_vision.py @@ -273,8 +273,7 @@ def forward( vision_feature_select_strategy: Optional[str] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/bamba/modular_bamba.py b/src/transformers/models/bamba/modular_bamba.py index 0d42d6e14329..303c5feeaf83 100644 --- a/src/transformers/models/bamba/modular_bamba.py +++ b/src/transformers/models/bamba/modular_bamba.py @@ -874,8 +874,7 @@ def forward( past_key_values: Optional[HybridMambaAttentionDynamicCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[BambaFlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -1109,8 +1108,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, diff --git a/src/transformers/models/bark/modeling_bark.py b/src/transformers/models/bark/modeling_bark.py index 4ee608d9aec6..883d88b80fda 100644 --- a/src/transformers/models/bark/modeling_bark.py +++ b/src/transformers/models/bark/modeling_bark.py @@ -513,8 +513,7 @@ def forward( labels: Optional[torch.LongTensor] = None, input_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithPast]: r""" @@ -1176,8 +1175,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, input_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" diff --git a/src/transformers/models/bart/modeling_bart.py b/src/transformers/models/bart/modeling_bart.py index f0adc76924fe..7e3156501dee 100755 --- a/src/transformers/models/bart/modeling_bart.py +++ b/src/transformers/models/bart/modeling_bart.py @@ -779,8 +779,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -959,8 +958,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: @@ -1258,8 +1256,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqModelOutput]: @@ -1431,8 +1428,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: @@ -1617,8 +1613,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: @@ -1763,8 +1758,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqQuestionAnsweringModelOutput]: @@ -1929,8 +1923,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/beit/modeling_beit.py b/src/transformers/models/beit/modeling_beit.py index 70a6c78e0651..78f113222006 100755 --- a/src/transformers/models/beit/modeling_beit.py +++ b/src/transformers/models/beit/modeling_beit.py @@ -809,8 +809,7 @@ def forward( pixel_values: torch.Tensor, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BeitModelOutputWithPooling]: @@ -907,8 +906,7 @@ def forward( bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: @@ -1002,8 +1000,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: @@ -1355,8 +1352,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, SemanticSegmenterOutput]: diff --git a/src/transformers/models/bert/modeling_bert.py b/src/transformers/models/bert/modeling_bert.py index 12080dfff6ff..62c4f41c0f25 100755 --- a/src/transformers/models/bert/modeling_bert.py +++ b/src/transformers/models/bert/modeling_bert.py @@ -910,8 +910,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1077,8 +1076,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, next_sentence_label: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BertForPreTrainingOutput]: r""" @@ -1187,8 +1185,7 @@ def forward( labels: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.Tensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **loss_kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1285,8 +1282,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1381,8 +1377,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], NextSentencePredictorOutput]: @@ -1488,8 +1483,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1576,8 +1570,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1684,8 +1677,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1751,8 +1743,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/bert_generation/modeling_bert_generation.py b/src/transformers/models/bert_generation/modeling_bert_generation.py index 959a3cce077d..ceab708c0dbc 100755 --- a/src/transformers/models/bert_generation/modeling_bert_generation.py +++ b/src/transformers/models/bert_generation/modeling_bert_generation.py @@ -660,8 +660,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: @@ -810,8 +809,7 @@ def forward( labels: Optional[torch.Tensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/big_bird/modeling_big_bird.py b/src/transformers/models/big_bird/modeling_big_bird.py index e06c8f87d5f8..b65d2067adc5 100755 --- a/src/transformers/models/big_bird/modeling_big_bird.py +++ b/src/transformers/models/big_bird/modeling_big_bird.py @@ -1902,8 +1902,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[BaseModelOutputWithPoolingAndCrossAttentions, tuple[torch.FloatTensor]]: @@ -2178,8 +2177,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.FloatTensor] = None, next_sentence_label: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BigBirdForPreTrainingOutput, tuple[torch.FloatTensor]]: r""" @@ -2287,8 +2285,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MaskedLMOutput, tuple[torch.FloatTensor]]: r""" @@ -2430,8 +2427,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[CausalLMOutputWithCrossAttentions, tuple[torch.FloatTensor]]: @@ -2546,8 +2542,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.FloatTensor]]: r""" @@ -2665,8 +2660,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MultipleChoiceModelOutput, tuple[torch.FloatTensor]]: r""" @@ -2773,8 +2767,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[TokenClassifierOutput, tuple[torch.FloatTensor]]: r""" @@ -2866,8 +2859,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BigBirdForQuestionAnsweringModelOutput, tuple[torch.FloatTensor]]: r""" diff --git a/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py b/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py index bc72d16bf547..737c3df2bb5d 100755 --- a/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py +++ b/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py @@ -1817,8 +1817,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): r""" @@ -2123,8 +2122,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ): @@ -2415,8 +2413,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqModelOutput]: @@ -2578,8 +2575,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: @@ -2735,8 +2731,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: @@ -2869,8 +2864,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqQuestionAnsweringModelOutput]: @@ -3019,8 +3013,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/biogpt/modular_biogpt.py b/src/transformers/models/biogpt/modular_biogpt.py index f37e15e0f079..22c0549599d5 100644 --- a/src/transformers/models/biogpt/modular_biogpt.py +++ b/src/transformers/models/biogpt/modular_biogpt.py @@ -368,8 +368,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], @@ -568,8 +567,7 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, **kwargs: Unpack[KwargsForCausalLM], @@ -660,8 +658,7 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, TokenClassifierOutput]: @@ -752,8 +749,7 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: diff --git a/src/transformers/models/blenderbot/modeling_blenderbot.py b/src/transformers/models/blenderbot/modeling_blenderbot.py index 0d699d01b59c..a8c41689cad8 100755 --- a/src/transformers/models/blenderbot/modeling_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_blenderbot.py @@ -1214,8 +1214,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: @@ -1392,8 +1391,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: @@ -1581,8 +1579,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py index 99666356d091..f8f57f85bf8a 100755 --- a/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py @@ -1181,8 +1181,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: @@ -1344,8 +1343,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: @@ -1533,8 +1531,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/blip/modeling_blip.py b/src/transformers/models/blip/modeling_blip.py index 3967f8bce478..d2c7d143a913 100644 --- a/src/transformers/models/blip/modeling_blip.py +++ b/src/transformers/models/blip/modeling_blip.py @@ -512,8 +512,7 @@ def forward( self, inputs_embeds, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -590,8 +589,7 @@ def __init__(self, config: BlipVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -826,8 +824,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BlipOutput]: @@ -953,8 +950,7 @@ def forward( pixel_values: torch.FloatTensor, input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, @@ -1135,8 +1131,7 @@ def forward( decoder_input_ids: Optional[torch.LongTensor] = None, decoder_attention_mask: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, @@ -1377,8 +1372,7 @@ def forward( pixel_values: torch.FloatTensor, use_itm_head: Optional[bool] = True, attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BlipTextVisionModelOutput]: diff --git a/src/transformers/models/blip/modeling_blip_text.py b/src/transformers/models/blip/modeling_blip_text.py index 151caef0b102..30c7a06beaea 100644 --- a/src/transformers/models/blip/modeling_blip_text.py +++ b/src/transformers/models/blip/modeling_blip_text.py @@ -667,8 +667,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, is_decoder: Optional[bool] = False, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: @@ -829,8 +828,7 @@ def forward( labels: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.Tensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, return_logits: Optional[bool] = False, is_decoder: Optional[bool] = True, diff --git a/src/transformers/models/blip_2/modeling_blip_2.py b/src/transformers/models/blip_2/modeling_blip_2.py index 970ee7ad0c44..3d47b4333e85 100644 --- a/src/transformers/models/blip_2/modeling_blip_2.py +++ b/src/transformers/models/blip_2/modeling_blip_2.py @@ -491,8 +491,7 @@ def forward( self, inputs_embeds, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -571,8 +570,7 @@ def __init__(self, config: Blip2VisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -1169,8 +1167,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -1339,8 +1336,7 @@ def get_text_features( decoder_input_ids: Optional[torch.Tensor] = None, decoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs: Unpack[KwargsForCausalLM], ): @@ -1414,8 +1410,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ): @@ -1460,8 +1455,7 @@ def get_image_features( def get_qformer_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ): @@ -1525,8 +1519,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, decoder_input_ids: Optional[torch.LongTensor] = None, decoder_attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, @@ -1694,8 +1687,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Blip2TextModelOutput]: r""" @@ -1782,8 +1774,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Blip2VisionModelOutput]: r""" @@ -1997,8 +1988,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, decoder_input_ids: Optional[torch.LongTensor] = None, decoder_attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, @@ -2305,8 +2295,7 @@ def forward( input_ids: torch.LongTensor, attention_mask: Optional[torch.LongTensor] = None, use_image_text_matching_head: Optional[bool] = False, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Blip2ImageTextMatchingModelOutput]: r""" diff --git a/src/transformers/models/bloom/modeling_bloom.py b/src/transformers/models/bloom/modeling_bloom.py index 66dfc0c1fa8c..8b07cc80cbc4 100644 --- a/src/transformers/models/bloom/modeling_bloom.py +++ b/src/transformers/models/bloom/modeling_bloom.py @@ -511,8 +511,7 @@ def forward( head_mask: Optional[torch.LongTensor] = None, inputs_embeds: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **deprecated_arguments, @@ -879,8 +878,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **deprecated_arguments, @@ -1014,8 +1012,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **deprecated_arguments, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutputWithPast]: @@ -1150,8 +1147,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **deprecated_arguments, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: @@ -1242,8 +1238,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/bridgetower/modeling_bridgetower.py b/src/transformers/models/bridgetower/modeling_bridgetower.py index 10db0bbb62fb..8bc56cd7c922 100644 --- a/src/transformers/models/bridgetower/modeling_bridgetower.py +++ b/src/transformers/models/bridgetower/modeling_bridgetower.py @@ -1060,8 +1060,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1239,8 +1238,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, image_token_type_idx: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, interpolate_pos_encoding: bool = False, @@ -1548,8 +1546,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[MaskedLMOutput, tuple[torch.FloatTensor]]: @@ -1650,8 +1647,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.FloatTensor]]: diff --git a/src/transformers/models/bros/modeling_bros.py b/src/transformers/models/bros/modeling_bros.py index 94d3a9d985d9..bcadbdf174cf 100755 --- a/src/transformers/models/bros/modeling_bros.py +++ b/src/transformers/models/bros/modeling_bros.py @@ -729,8 +729,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -890,8 +889,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1015,8 +1013,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, initial_token_labels: Optional[torch.Tensor] = None, subsequent_token_labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BrosSpadeOutput]: r""" @@ -1156,8 +1153,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" diff --git a/src/transformers/models/camembert/modeling_camembert.py b/src/transformers/models/camembert/modeling_camembert.py index 1b4a52295f28..94344162333c 100644 --- a/src/transformers/models/camembert/modeling_camembert.py +++ b/src/transformers/models/camembert/modeling_camembert.py @@ -823,8 +823,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -992,8 +991,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1077,8 +1075,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1172,8 +1169,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1283,8 +1279,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1362,8 +1357,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" @@ -1469,8 +1463,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/canine/modeling_canine.py b/src/transformers/models/canine/modeling_canine.py index d55c600d05de..8bd1a5c88477 100644 --- a/src/transformers/models/canine/modeling_canine.py +++ b/src/transformers/models/canine/modeling_canine.py @@ -1017,8 +1017,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, CanineModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1196,8 +1195,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1281,8 +1279,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1386,8 +1383,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1483,8 +1479,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/chameleon/modeling_chameleon.py b/src/transformers/models/chameleon/modeling_chameleon.py index d6575a8751d0..9053e91a30c0 100644 --- a/src/transformers/models/chameleon/modeling_chameleon.py +++ b/src/transformers/models/chameleon/modeling_chameleon.py @@ -944,8 +944,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], diff --git a/src/transformers/models/chinese_clip/modeling_chinese_clip.py b/src/transformers/models/chinese_clip/modeling_chinese_clip.py index 5de98397cad0..79ba8a059a5b 100644 --- a/src/transformers/models/chinese_clip/modeling_chinese_clip.py +++ b/src/transformers/models/chinese_clip/modeling_chinese_clip.py @@ -888,8 +888,7 @@ def __init__(self, config: ChineseCLIPConfig): def forward( self, inputs_embeds, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -962,8 +961,7 @@ def __init__(self, config: ChineseCLIPVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -1065,8 +1063,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1187,8 +1184,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -1266,8 +1262,7 @@ def get_text_features( attention_mask: Optional[torch.Tensor] = None, token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1313,8 +1308,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> torch.FloatTensor: @@ -1370,8 +1364,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, ChineseCLIPOutput]: diff --git a/src/transformers/models/clap/modeling_clap.py b/src/transformers/models/clap/modeling_clap.py index 0f8058ad76ac..9b4e80a81449 100644 --- a/src/transformers/models/clap/modeling_clap.py +++ b/src/transformers/models/clap/modeling_clap.py @@ -1598,8 +1598,7 @@ def forward( self, input_features: Optional[torch.FloatTensor] = None, is_longer: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1694,8 +1693,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1835,8 +1833,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1883,8 +1880,7 @@ def get_audio_features( input_features: Optional[torch.Tensor] = None, is_longer: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1939,8 +1935,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ClapOutput]: r""" @@ -2057,8 +2052,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ClapTextModelOutput]: r""" @@ -2122,8 +2116,7 @@ def forward( self, input_features: Optional[torch.FloatTensor] = None, is_longer: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ClapAudioModelOutput]: r""" diff --git a/src/transformers/models/clip/modeling_clip.py b/src/transformers/models/clip/modeling_clip.py index b93f63bcea97..bf6bf85b5b30 100644 --- a/src/transformers/models/clip/modeling_clip.py +++ b/src/transformers/models/clip/modeling_clip.py @@ -531,8 +531,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutput: r""" Args: @@ -628,8 +627,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -725,8 +723,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: r""" Examples: @@ -769,8 +766,7 @@ def __init__(self, config: CLIPVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = False, ) -> BaseModelOutputWithPooling: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -826,8 +822,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> BaseModelOutputWithPooling: r""" @@ -905,8 +900,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> torch.FloatTensor: r""" Returns: @@ -947,8 +941,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> torch.FloatTensor: r""" @@ -1000,8 +993,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> CLIPOutput: r""" @@ -1111,8 +1103,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> CLIPTextModelOutput: r""" Examples: @@ -1171,8 +1162,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> CLIPVisionModelOutput: r""" @@ -1242,8 +1232,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> ImageClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): diff --git a/src/transformers/models/clipseg/modeling_clipseg.py b/src/transformers/models/clipseg/modeling_clipseg.py index c68404cb66c8..6b81fdafe563 100644 --- a/src/transformers/models/clipseg/modeling_clipseg.py +++ b/src/transformers/models/clipseg/modeling_clipseg.py @@ -493,8 +493,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -588,8 +587,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -684,8 +682,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -729,8 +726,7 @@ def __init__(self, config: CLIPSegVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor], - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = True, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -782,8 +778,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = True, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -858,8 +853,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -903,8 +897,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = True, return_dict: Optional[bool] = None, ) -> torch.FloatTensor: @@ -958,8 +951,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = True, return_dict: Optional[bool] = None, ) -> Union[tuple, CLIPSegOutput]: @@ -1150,8 +1142,7 @@ def forward( self, hidden_states: tuple[torch.Tensor], conditional_embeddings: torch.Tensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = True, ): all_hidden_states = () if output_hidden_states else None @@ -1263,8 +1254,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = True, return_dict: Optional[bool] = None, ) -> Union[tuple, CLIPSegOutput]: diff --git a/src/transformers/models/clvp/modeling_clvp.py b/src/transformers/models/clvp/modeling_clvp.py index a2e42fc2aea8..3d072cea61a6 100644 --- a/src/transformers/models/clvp/modeling_clvp.py +++ b/src/transformers/models/clvp/modeling_clvp.py @@ -881,8 +881,7 @@ def forward( inputs_embeds: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.LongTensor] = None, position_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1045,8 +1044,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1208,8 +1206,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1402,8 +1399,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: r""" diff --git a/src/transformers/models/codegen/modeling_codegen.py b/src/transformers/models/codegen/modeling_codegen.py index 6a99d0fa390a..9401f419ac12 100644 --- a/src/transformers/models/codegen/modeling_codegen.py +++ b/src/transformers/models/codegen/modeling_codegen.py @@ -357,8 +357,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwargs, for now @@ -651,8 +650,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs, diff --git a/src/transformers/models/cohere/modular_cohere.py b/src/transformers/models/cohere/modular_cohere.py index 248ada3227f4..c5e5b957ec99 100644 --- a/src/transformers/models/cohere/modular_cohere.py +++ b/src/transformers/models/cohere/modular_cohere.py @@ -315,8 +315,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], diff --git a/src/transformers/models/cohere2/modular_cohere2.py b/src/transformers/models/cohere2/modular_cohere2.py index f14ffc85fdfa..5838b09e19b0 100644 --- a/src/transformers/models/cohere2/modular_cohere2.py +++ b/src/transformers/models/cohere2/modular_cohere2.py @@ -412,8 +412,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: diff --git a/src/transformers/models/colpali/modeling_colpali.py b/src/transformers/models/colpali/modeling_colpali.py index 077ec0a88de5..b108bea2adfd 100644 --- a/src/transformers/models/colpali/modeling_colpali.py +++ b/src/transformers/models/colpali/modeling_colpali.py @@ -130,8 +130,7 @@ def forward( input_ids: Optional[torch.LongTensor] = None, pixel_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> ColPaliForRetrievalOutput: diff --git a/src/transformers/models/conditional_detr/modeling_conditional_detr.py b/src/transformers/models/conditional_detr/modeling_conditional_detr.py index 19b1439302eb..d6a83cf5a8fb 100644 --- a/src/transformers/models/conditional_detr/modeling_conditional_detr.py +++ b/src/transformers/models/conditional_detr/modeling_conditional_detr.py @@ -1418,8 +1418,7 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], ConditionalDetrModelOutput]: r""" diff --git a/src/transformers/models/convbert/modeling_convbert.py b/src/transformers/models/convbert/modeling_convbert.py index 1a443c575ab0..7ef942ea3058 100755 --- a/src/transformers/models/convbert/modeling_convbert.py +++ b/src/transformers/models/convbert/modeling_convbert.py @@ -816,8 +816,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -919,8 +918,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1016,8 +1014,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1100,8 +1097,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1209,8 +1205,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1276,8 +1271,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/cpmant/modeling_cpmant.py b/src/transformers/models/cpmant/modeling_cpmant.py index f3ecf4930f6e..5eac5b6dcea3 100755 --- a/src/transformers/models/cpmant/modeling_cpmant.py +++ b/src/transformers/models/cpmant/modeling_cpmant.py @@ -714,8 +714,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, past_key_values: Optional[list[tuple[torch.Tensor, torch.Tensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, attention_mask: Optional[torch.Tensor] = None, # dummy parameter for text-generation pipeline diff --git a/src/transformers/models/csm/modular_csm.py b/src/transformers/models/csm/modular_csm.py index 7f882ca2c0ec..4ec15b5d3dcd 100644 --- a/src/transformers/models/csm/modular_csm.py +++ b/src/transformers/models/csm/modular_csm.py @@ -184,8 +184,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: @@ -375,8 +374,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -685,8 +683,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], diff --git a/src/transformers/models/ctrl/modeling_ctrl.py b/src/transformers/models/ctrl/modeling_ctrl.py index 883803741d7e..b6c33789ad2f 100644 --- a/src/transformers/models/ctrl/modeling_ctrl.py +++ b/src/transformers/models/ctrl/modeling_ctrl.py @@ -273,8 +273,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPast]: @@ -459,8 +458,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithPast]: @@ -613,8 +611,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/dab_detr/modeling_dab_detr.py b/src/transformers/models/dab_detr/modeling_dab_detr.py index a9369b8700d4..b1642f88f9fb 100644 --- a/src/transformers/models/dab_detr/modeling_dab_detr.py +++ b/src/transformers/models/dab_detr/modeling_dab_detr.py @@ -923,8 +923,7 @@ def forward( inputs_embeds, attention_mask, object_queries, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): r""" @@ -1062,8 +1061,7 @@ def forward( memory_key_padding_mask, object_queries, query_position_embeddings, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): r""" @@ -1287,8 +1285,7 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DabDetrModelOutput]: r""" diff --git a/src/transformers/models/data2vec/modeling_data2vec_audio.py b/src/transformers/models/data2vec/modeling_data2vec_audio.py index 940af308cd78..7605c9f980c2 100755 --- a/src/transformers/models/data2vec/modeling_data2vec_audio.py +++ b/src/transformers/models/data2vec/modeling_data2vec_audio.py @@ -822,8 +822,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Data2VecAudioBaseModelOutput]: r""" @@ -935,8 +934,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -1058,8 +1056,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: @@ -1173,8 +1170,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1354,8 +1350,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, XVectorOutput]: diff --git a/src/transformers/models/data2vec/modeling_data2vec_text.py b/src/transformers/models/data2vec/modeling_data2vec_text.py index 0d7e85134907..612d2d80fb17 100644 --- a/src/transformers/models/data2vec/modeling_data2vec_text.py +++ b/src/transformers/models/data2vec/modeling_data2vec_text.py @@ -658,8 +658,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -797,8 +796,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -915,8 +913,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1022,8 +1019,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1107,8 +1103,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1216,8 +1211,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1308,8 +1302,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/data2vec/modeling_data2vec_vision.py b/src/transformers/models/data2vec/modeling_data2vec_vision.py index 8762fedb1021..c5881c7b39aa 100644 --- a/src/transformers/models/data2vec/modeling_data2vec_vision.py +++ b/src/transformers/models/data2vec/modeling_data2vec_vision.py @@ -815,8 +815,7 @@ def forward( pixel_values: torch.Tensor, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, Data2VecVisionModelOutputWithPooling]: @@ -911,8 +910,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: @@ -1276,8 +1274,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, SemanticSegmenterOutput]: diff --git a/src/transformers/models/dbrx/modeling_dbrx.py b/src/transformers/models/dbrx/modeling_dbrx.py index 097370a46cd5..29d9b4501ad6 100644 --- a/src/transformers/models/dbrx/modeling_dbrx.py +++ b/src/transformers/models/dbrx/modeling_dbrx.py @@ -871,8 +871,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1173,8 +1172,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/deberta/modeling_deberta.py b/src/transformers/models/deberta/modeling_deberta.py index eef11f7ec34b..a77aa106a8f0 100644 --- a/src/transformers/models/deberta/modeling_deberta.py +++ b/src/transformers/models/deberta/modeling_deberta.py @@ -680,8 +680,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -886,8 +885,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -995,8 +993,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1089,8 +1086,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1151,8 +1147,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/deberta_v2/modeling_deberta_v2.py b/src/transformers/models/deberta_v2/modeling_deberta_v2.py index 5073d1de5263..38674e1cf563 100644 --- a/src/transformers/models/deberta_v2/modeling_deberta_v2.py +++ b/src/transformers/models/deberta_v2/modeling_deberta_v2.py @@ -758,8 +758,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -963,8 +962,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1074,8 +1072,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1169,8 +1166,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1232,8 +1228,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1320,8 +1315,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" diff --git a/src/transformers/models/decision_transformer/modeling_decision_transformer.py b/src/transformers/models/decision_transformer/modeling_decision_transformer.py index d34e989986d1..f24ad353cd30 100755 --- a/src/transformers/models/decision_transformer/modeling_decision_transformer.py +++ b/src/transformers/models/decision_transformer/modeling_decision_transformer.py @@ -527,8 +527,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions diff --git a/src/transformers/models/deformable_detr/modeling_deformable_detr.py b/src/transformers/models/deformable_detr/modeling_deformable_detr.py index e36da6da89bc..4f7fb40ff2c0 100755 --- a/src/transformers/models/deformable_detr/modeling_deformable_detr.py +++ b/src/transformers/models/deformable_detr/modeling_deformable_detr.py @@ -1544,8 +1544,7 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DeformableDetrModelOutput]: r""" diff --git a/src/transformers/models/deit/modeling_deit.py b/src/transformers/models/deit/modeling_deit.py index 6b6284995fc7..a732b8609dbc 100644 --- a/src/transformers/models/deit/modeling_deit.py +++ b/src/transformers/models/deit/modeling_deit.py @@ -512,8 +512,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -621,8 +620,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, MaskedImageModelingOutput]: @@ -726,8 +724,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, ImageClassifierOutput]: @@ -878,8 +875,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, DeiTForImageClassificationWithTeacherOutput]: diff --git a/src/transformers/models/depth_anything/modeling_depth_anything.py b/src/transformers/models/depth_anything/modeling_depth_anything.py index f5ace462120a..054eceb77d06 100644 --- a/src/transformers/models/depth_anything/modeling_depth_anything.py +++ b/src/transformers/models/depth_anything/modeling_depth_anything.py @@ -346,8 +346,7 @@ def forward( self, pixel_values: torch.FloatTensor, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], DepthEstimatorOutput]: r""" diff --git a/src/transformers/models/depth_pro/modeling_depth_pro.py b/src/transformers/models/depth_pro/modeling_depth_pro.py index e9d513991aa5..d0bee293b0a6 100644 --- a/src/transformers/models/depth_pro/modeling_depth_pro.py +++ b/src/transformers/models/depth_pro/modeling_depth_pro.py @@ -670,8 +670,7 @@ def forward( self, pixel_values: torch.FloatTensor, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, DepthProOutput]: r""" @@ -1057,8 +1056,7 @@ def forward( pixel_values: torch.FloatTensor, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], DepthProDepthEstimatorOutput]: r""" diff --git a/src/transformers/models/detr/modeling_detr.py b/src/transformers/models/detr/modeling_detr.py index 9f8ea167ab7b..e88667a72d4a 100644 --- a/src/transformers/models/detr/modeling_detr.py +++ b/src/transformers/models/detr/modeling_detr.py @@ -1152,8 +1152,7 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DetrModelOutput]: r""" diff --git a/src/transformers/models/diffllama/modeling_diffllama.py b/src/transformers/models/diffllama/modeling_diffllama.py index 25732f685943..be314d808860 100644 --- a/src/transformers/models/diffllama/modeling_diffllama.py +++ b/src/transformers/models/diffllama/modeling_diffllama.py @@ -834,8 +834,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -923,8 +922,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.transformer( @@ -992,8 +990,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/dinat/modeling_dinat.py b/src/transformers/models/dinat/modeling_dinat.py index 25a1a3ae5c6b..e598ffd8b152 100644 --- a/src/transformers/models/dinat/modeling_dinat.py +++ b/src/transformers/models/dinat/modeling_dinat.py @@ -662,8 +662,7 @@ class PreTrainedModel def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, DinatModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -734,8 +733,7 @@ def forward( self, pixel_values: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, DinatImageClassifierOutput]: r""" diff --git a/src/transformers/models/dinov2/modeling_dinov2.py b/src/transformers/models/dinov2/modeling_dinov2.py index 8dfd7fc40975..cace1ce316ad 100644 --- a/src/transformers/models/dinov2/modeling_dinov2.py +++ b/src/transformers/models/dinov2/modeling_dinov2.py @@ -560,8 +560,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -637,8 +636,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: r""" diff --git a/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py b/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py index 5145221e0366..48d425bfd4a0 100644 --- a/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py +++ b/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py @@ -320,8 +320,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: r""" diff --git a/src/transformers/models/distilbert/modeling_distilbert.py b/src/transformers/models/distilbert/modeling_distilbert.py index 1a84544bee3f..c5c1bdd0b3c0 100755 --- a/src/transformers/models/distilbert/modeling_distilbert.py +++ b/src/transformers/models/distilbert/modeling_distilbert.py @@ -684,8 +684,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BaseModelOutput, tuple[torch.Tensor, ...]]: r""" @@ -803,8 +802,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MaskedLMOutput, tuple[torch.Tensor, ...]]: r""" @@ -905,8 +903,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.Tensor, ...]]: r""" @@ -1012,8 +1009,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[QuestionAnsweringModelOutput, tuple[torch.Tensor, ...]]: r""" @@ -1119,8 +1115,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[TokenClassifierOutput, tuple[torch.Tensor, ...]]: r""" @@ -1202,8 +1197,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MultipleChoiceModelOutput, tuple[torch.Tensor, ...]]: r""" diff --git a/src/transformers/models/donut/modeling_donut_swin.py b/src/transformers/models/donut/modeling_donut_swin.py index 222b09864230..228606bca90f 100644 --- a/src/transformers/models/donut/modeling_donut_swin.py +++ b/src/transformers/models/donut/modeling_donut_swin.py @@ -936,8 +936,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, DonutSwinModelOutput]: @@ -1031,8 +1030,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, DonutSwinImageClassifierOutput]: diff --git a/src/transformers/models/dpr/modeling_dpr.py b/src/transformers/models/dpr/modeling_dpr.py index 0dfe85f4e060..43415a5ed28a 100644 --- a/src/transformers/models/dpr/modeling_dpr.py +++ b/src/transformers/models/dpr/modeling_dpr.py @@ -325,8 +325,7 @@ def forward( attention_mask: Optional[Tensor] = None, token_type_ids: Optional[Tensor] = None, inputs_embeds: Optional[Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[DPRContextEncoderOutput, tuple[Tensor, ...]]: r""" @@ -430,8 +429,7 @@ def forward( attention_mask: Optional[Tensor] = None, token_type_ids: Optional[Tensor] = None, inputs_embeds: Optional[Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[DPRQuestionEncoderOutput, tuple[Tensor, ...]]: r""" @@ -535,8 +533,7 @@ def forward( input_ids: Optional[Tensor] = None, attention_mask: Optional[Tensor] = None, inputs_embeds: Optional[Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[DPRReaderOutput, tuple[Tensor, ...]]: r""" diff --git a/src/transformers/models/dpt/modeling_dpt.py b/src/transformers/models/dpt/modeling_dpt.py index d3c1703b9ee7..135a8d6ef3f8 100755 --- a/src/transformers/models/dpt/modeling_dpt.py +++ b/src/transformers/models/dpt/modeling_dpt.py @@ -887,8 +887,7 @@ def forward( self, pixel_values: torch.FloatTensor, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndIntermediateActivations]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1073,8 +1072,7 @@ def forward( pixel_values: torch.FloatTensor, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], DepthEstimatorOutput]: r""" @@ -1246,8 +1244,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SemanticSegmenterOutput]: r""" diff --git a/src/transformers/models/electra/modeling_electra.py b/src/transformers/models/electra/modeling_electra.py index dfe5849a5a7d..eab5c1a9fd50 100644 --- a/src/transformers/models/electra/modeling_electra.py +++ b/src/transformers/models/electra/modeling_electra.py @@ -747,8 +747,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -975,8 +974,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1063,8 +1061,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], ElectraForPreTrainingOutput]: r""" @@ -1175,8 +1172,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1252,8 +1248,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1321,8 +1316,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1400,8 +1394,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1522,8 +1515,7 @@ def forward( labels: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.Tensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/emu3/modular_emu3.py b/src/transformers/models/emu3/modular_emu3.py index ed58e43f86fa..c1b285317fb3 100644 --- a/src/transformers/models/emu3/modular_emu3.py +++ b/src/transformers/models/emu3/modular_emu3.py @@ -1004,8 +1004,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1119,8 +1118,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/encoder_decoder/modeling_encoder_decoder.py b/src/transformers/models/encoder_decoder/modeling_encoder_decoder.py index 796a104023db..e21339fb6f0c 100644 --- a/src/transformers/models/encoder_decoder/modeling_encoder_decoder.py +++ b/src/transformers/models/encoder_decoder/modeling_encoder_decoder.py @@ -458,8 +458,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Seq2SeqLMOutput]: diff --git a/src/transformers/models/ernie/modeling_ernie.py b/src/transformers/models/ernie/modeling_ernie.py index 21a79a3fabc0..18cf43b5f2d3 100644 --- a/src/transformers/models/ernie/modeling_ernie.py +++ b/src/transformers/models/ernie/modeling_ernie.py @@ -752,8 +752,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -899,8 +898,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, next_sentence_label: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], ErnieForPreTrainingOutput]: r""" @@ -1019,8 +1017,7 @@ def forward( labels: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.Tensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1133,8 +1130,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1238,8 +1234,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], NextSentencePredictorOutput]: @@ -1353,8 +1348,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1449,8 +1443,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1565,8 +1558,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1640,8 +1632,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/esm/modeling_esm.py b/src/transformers/models/esm/modeling_esm.py index dbf260fb2160..cc0c8f761706 100755 --- a/src/transformers/models/esm/modeling_esm.py +++ b/src/transformers/models/esm/modeling_esm.py @@ -881,8 +881,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -1042,8 +1041,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1140,8 +1138,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1225,8 +1222,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" diff --git a/src/transformers/models/falcon/modeling_falcon.py b/src/transformers/models/falcon/modeling_falcon.py index 455afbb21577..f7a6d9398307 100644 --- a/src/transformers/models/falcon/modeling_falcon.py +++ b/src/transformers/models/falcon/modeling_falcon.py @@ -735,8 +735,7 @@ def forward( head_mask: Optional[torch.LongTensor] = None, inputs_embeds: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor, ...], BaseModelOutputWithPastAndCrossAttentions]: @@ -1065,8 +1064,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, @@ -1190,8 +1188,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutputWithPast]: r""" @@ -1316,8 +1313,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1395,8 +1391,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/falcon_h1/modular_falcon_h1.py b/src/transformers/models/falcon_h1/modular_falcon_h1.py index 85cd4db9116a..14a89f2afe95 100644 --- a/src/transformers/models/falcon_h1/modular_falcon_h1.py +++ b/src/transformers/models/falcon_h1/modular_falcon_h1.py @@ -1039,8 +1039,7 @@ def forward( past_key_values: Optional[FalconHybridMambaAttentionDynamicCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, BaseModelOutputWithPast]: @@ -1264,8 +1263,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, diff --git a/src/transformers/models/flaubert/modeling_flaubert.py b/src/transformers/models/flaubert/modeling_flaubert.py index e97f91a11130..f1019480b56d 100644 --- a/src/transformers/models/flaubert/modeling_flaubert.py +++ b/src/transformers/models/flaubert/modeling_flaubert.py @@ -810,8 +810,7 @@ def forward( cache: Optional[dict[str, torch.FloatTensor]] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1035,8 +1034,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1125,8 +1123,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1236,8 +1233,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1333,8 +1329,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" @@ -1485,8 +1480,7 @@ def forward( is_impossible: Optional[torch.Tensor] = None, cls_index: Optional[torch.Tensor] = None, p_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, FlaubertForQuestionAnsweringOutput]: r""" @@ -1606,8 +1600,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" diff --git a/src/transformers/models/flava/modeling_flava.py b/src/transformers/models/flava/modeling_flava.py index 99ffbaa9f307..e372f521bf23 100644 --- a/src/transformers/models/flava/modeling_flava.py +++ b/src/transformers/models/flava/modeling_flava.py @@ -770,8 +770,7 @@ def forward( interpolate_pos_encoding: Optional[bool] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -865,8 +864,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -973,8 +971,7 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1084,8 +1081,7 @@ def get_text_features( attention_mask: Optional[torch.Tensor] = None, token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1141,8 +1137,7 @@ def get_image_features( interpolate_pos_encoding: Optional[bool] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" diff --git a/src/transformers/models/fsmt/modeling_fsmt.py b/src/transformers/models/fsmt/modeling_fsmt.py index 9e3000ad1338..d3830b85b216 100644 --- a/src/transformers/models/fsmt/modeling_fsmt.py +++ b/src/transformers/models/fsmt/modeling_fsmt.py @@ -948,8 +948,7 @@ def forward( encoder_outputs: Optional[tuple[torch.FloatTensor]] = None, past_key_values: Optional[tuple[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, return_dict: Optional[bool] = None, @@ -1095,8 +1094,7 @@ def forward( decoder_inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqLMOutput]: r""" diff --git a/src/transformers/models/funnel/modeling_funnel.py b/src/transformers/models/funnel/modeling_funnel.py index bb239566c0bc..a134ab9c7f13 100644 --- a/src/transformers/models/funnel/modeling_funnel.py +++ b/src/transformers/models/funnel/modeling_funnel.py @@ -863,8 +863,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -930,8 +929,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1022,8 +1020,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, FunnelForPreTrainingOutput]: r""" @@ -1111,8 +1108,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1178,8 +1174,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1257,8 +1252,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1332,8 +1326,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1394,8 +1387,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/gemma/modular_gemma.py b/src/transformers/models/gemma/modular_gemma.py index f20f2b7c9993..5532dbaf7a7f 100644 --- a/src/transformers/models/gemma/modular_gemma.py +++ b/src/transformers/models/gemma/modular_gemma.py @@ -377,8 +377,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: diff --git a/src/transformers/models/gemma2/modular_gemma2.py b/src/transformers/models/gemma2/modular_gemma2.py index 77c38ad5117b..2a2e43bfe0b2 100644 --- a/src/transformers/models/gemma2/modular_gemma2.py +++ b/src/transformers/models/gemma2/modular_gemma2.py @@ -378,8 +378,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -507,8 +506,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **loss_kwargs, diff --git a/src/transformers/models/gemma3/modular_gemma3.py b/src/transformers/models/gemma3/modular_gemma3.py index 3ce4bb4813c4..ac076fd26037 100644 --- a/src/transformers/models/gemma3/modular_gemma3.py +++ b/src/transformers/models/gemma3/modular_gemma3.py @@ -564,8 +564,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -775,8 +774,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **lm_kwargs, ) -> Union[tuple, Gemma3ModelOutputWithPast]: @@ -886,8 +884,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **lm_kwargs, diff --git a/src/transformers/models/got_ocr2/modular_got_ocr2.py b/src/transformers/models/got_ocr2/modular_got_ocr2.py index d8fdd7249f4f..32ef0abb9ce3 100644 --- a/src/transformers/models/got_ocr2/modular_got_ocr2.py +++ b/src/transformers/models/got_ocr2/modular_got_ocr2.py @@ -327,8 +327,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -398,8 +397,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/gpt2/modeling_gpt2.py b/src/transformers/models/gpt2/modeling_gpt2.py index 584d21c41087..30012b583748 100644 --- a/src/transformers/models/gpt2/modeling_gpt2.py +++ b/src/transformers/models/gpt2/modeling_gpt2.py @@ -785,8 +785,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: @@ -1182,8 +1181,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1335,8 +1333,7 @@ def forward( labels: Optional[torch.LongTensor] = None, mc_labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, GPT2DoubleHeadsModelOutput]: @@ -1499,8 +1496,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: r""" @@ -1632,8 +1628,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1718,8 +1713,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/gpt_neo/modeling_gpt_neo.py b/src/transformers/models/gpt_neo/modeling_gpt_neo.py index 417eb11ab0ff..c62cb3aaf524 100755 --- a/src/transformers/models/gpt_neo/modeling_gpt_neo.py +++ b/src/transformers/models/gpt_neo/modeling_gpt_neo.py @@ -542,8 +542,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: @@ -843,8 +842,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs, @@ -969,8 +967,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutputWithPast]: r""" @@ -1092,8 +1089,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1174,8 +1170,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/gpt_neox/modular_gpt_neox.py b/src/transformers/models/gpt_neox/modular_gpt_neox.py index 095188d21009..58af6ff0f2fd 100644 --- a/src/transformers/models/gpt_neox/modular_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modular_gpt_neox.py @@ -295,8 +295,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, past_key_values: Optional[Cache] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -446,8 +445,7 @@ def forward( past_key_values: Optional[Union[Cache, tuple[tuple[torch.FloatTensor]]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -543,8 +541,7 @@ def forward( past_key_values: Optional[Union[Cache, tuple[tuple[torch.FloatTensor]]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> SequenceClassifierOutputWithPast: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -624,8 +621,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> TokenClassifierOutput: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): @@ -685,8 +681,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.gpt_neox( input_ids, diff --git a/src/transformers/models/gptj/modeling_gptj.py b/src/transformers/models/gptj/modeling_gptj.py index 1e9c0ef5332f..59871c48377c 100644 --- a/src/transformers/models/gptj/modeling_gptj.py +++ b/src/transformers/models/gptj/modeling_gptj.py @@ -638,8 +638,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: @@ -986,8 +985,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs, @@ -1111,8 +1109,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: r""" @@ -1228,8 +1225,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/granite/modular_granite.py b/src/transformers/models/granite/modular_granite.py index b54d7eec6b00..6ab58b1c67f6 100644 --- a/src/transformers/models/granite/modular_granite.py +++ b/src/transformers/models/granite/modular_granite.py @@ -138,8 +138,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -240,8 +239,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], diff --git a/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py b/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py index b61c4ad61b87..4721dd673fbb 100644 --- a/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py +++ b/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py @@ -201,8 +201,7 @@ def forward( past_key_values: Optional[Union[Cache, list[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/groupvit/modeling_groupvit.py b/src/transformers/models/groupvit/modeling_groupvit.py index 100fd2dd85f5..d5e37cdc7f77 100644 --- a/src/transformers/models/groupvit/modeling_groupvit.py +++ b/src/transformers/models/groupvit/modeling_groupvit.py @@ -860,8 +860,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -955,8 +954,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1050,8 +1048,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1149,8 +1146,7 @@ def get_input_embeddings(self) -> GroupViTPatchEmbeddings: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1234,8 +1230,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1279,8 +1274,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1332,8 +1326,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_segmentation: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, GroupViTModelOutput]: diff --git a/src/transformers/models/hubert/modular_hubert.py b/src/transformers/models/hubert/modular_hubert.py index ec13ad609009..9de202cd6dd9 100644 --- a/src/transformers/models/hubert/modular_hubert.py +++ b/src/transformers/models/hubert/modular_hubert.py @@ -225,8 +225,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" diff --git a/src/transformers/models/ijepa/modular_ijepa.py b/src/transformers/models/ijepa/modular_ijepa.py index 2f6064acbb11..0673ea9d71ec 100644 --- a/src/transformers/models/ijepa/modular_ijepa.py +++ b/src/transformers/models/ijepa/modular_ijepa.py @@ -159,8 +159,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: diff --git a/src/transformers/models/informer/modular_informer.py b/src/transformers/models/informer/modular_informer.py index 755fcd68853a..db3d52564d5e 100644 --- a/src/transformers/models/informer/modular_informer.py +++ b/src/transformers/models/informer/modular_informer.py @@ -537,8 +537,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" diff --git a/src/transformers/models/instructblipvideo/modular_instructblipvideo.py b/src/transformers/models/instructblipvideo/modular_instructblipvideo.py index e941b34c9f67..87f4fcf72b9b 100644 --- a/src/transformers/models/instructblipvideo/modular_instructblipvideo.py +++ b/src/transformers/models/instructblipvideo/modular_instructblipvideo.py @@ -204,8 +204,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, decoder_input_ids: Optional[torch.LongTensor] = None, decoder_attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, use_cache: Optional[bool] = None, @@ -374,8 +373,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, decoder_input_ids: Optional[torch.LongTensor] = None, decoder_attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, diff --git a/src/transformers/models/internvl/modular_internvl.py b/src/transformers/models/internvl/modular_internvl.py index 7ba30c922740..2c567e28d86e 100644 --- a/src/transformers/models/internvl/modular_internvl.py +++ b/src/transformers/models/internvl/modular_internvl.py @@ -450,8 +450,7 @@ def forward( self, pixel_values: torch.Tensor, bool_masked_pos: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> Union[tuple, InternVLVisionModelOutputWithPooling]: r""" bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): diff --git a/src/transformers/models/janus/modular_janus.py b/src/transformers/models/janus/modular_janus.py index 5acbf9a0fbb5..4814dc4db239 100644 --- a/src/transformers/models/janus/modular_janus.py +++ b/src/transformers/models/janus/modular_janus.py @@ -932,8 +932,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, ): @@ -1045,8 +1044,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, ): diff --git a/src/transformers/models/layoutlm/modeling_layoutlm.py b/src/transformers/models/layoutlm/modeling_layoutlm.py index 6ce2e7e2dcf2..81eb5aaa94c3 100644 --- a/src/transformers/models/layoutlm/modeling_layoutlm.py +++ b/src/transformers/models/layoutlm/modeling_layoutlm.py @@ -671,8 +671,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -820,8 +819,7 @@ def forward( labels: Optional[torch.LongTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -943,8 +941,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1081,8 +1078,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1199,8 +1195,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py b/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py index fa89c6c45b41..fb91e3c8e3a4 100755 --- a/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py +++ b/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py @@ -736,8 +736,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -906,8 +905,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1111,8 +1109,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1264,8 +1261,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py b/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py index 83f87ec5281b..c6873d957e77 100644 --- a/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py +++ b/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py @@ -690,8 +690,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -938,8 +937,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, pixel_values: Optional[torch.LongTensor] = None, ) -> Union[tuple, TokenClassifierOutput]: @@ -1039,8 +1037,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, bbox: Optional[torch.LongTensor] = None, pixel_values: Optional[torch.LongTensor] = None, @@ -1159,8 +1156,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, bbox: Optional[torch.LongTensor] = None, pixel_values: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/led/modeling_led.py b/src/transformers/models/led/modeling_led.py index bc5738613218..c840e85ead88 100755 --- a/src/transformers/models/led/modeling_led.py +++ b/src/transformers/models/led/modeling_led.py @@ -2046,8 +2046,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], LEDSeq2SeqModelOutput]: r""" @@ -2216,8 +2215,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], LEDSeq2SeqLMOutput]: r""" @@ -2422,8 +2420,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], LEDSeq2SeqSequenceClassifierOutput]: r""" @@ -2574,8 +2571,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], LEDSeq2SeqQuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/lilt/modeling_lilt.py b/src/transformers/models/lilt/modeling_lilt.py index 0c76a25a6e3d..24a35ccfce1b 100644 --- a/src/transformers/models/lilt/modeling_lilt.py +++ b/src/transformers/models/lilt/modeling_lilt.py @@ -633,8 +633,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPooling]: r""" @@ -769,8 +768,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -886,8 +884,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1005,8 +1002,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index b3fc2b9c8424..c4bf565e78b6 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -571,8 +571,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -661,8 +660,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.transformer( @@ -730,8 +728,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/llava_next_video/modular_llava_next_video.py b/src/transformers/models/llava_next_video/modular_llava_next_video.py index 75aaa7ed7bbd..490632888bc6 100644 --- a/src/transformers/models/llava_next_video/modular_llava_next_video.py +++ b/src/transformers/models/llava_next_video/modular_llava_next_video.py @@ -392,8 +392,7 @@ def forward( vision_feature_layer: Optional[Union[int, list[int]]] = None, vision_feature_select_strategy: Optional[str] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -522,8 +521,7 @@ def forward( vision_feature_select_strategy: Optional[str] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/llava_onevision/modular_llava_onevision.py b/src/transformers/models/llava_onevision/modular_llava_onevision.py index c4ba5c0ed32e..ed476b6d93c4 100644 --- a/src/transformers/models/llava_onevision/modular_llava_onevision.py +++ b/src/transformers/models/llava_onevision/modular_llava_onevision.py @@ -488,8 +488,7 @@ def forward( vision_aspect_ratio: Optional[str] = None, batch_num_images: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -629,8 +628,7 @@ def forward( batch_num_images: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/longformer/modeling_longformer.py b/src/transformers/models/longformer/modeling_longformer.py index a40d5bb0e2c9..f1fec421e8ba 100755 --- a/src/transformers/models/longformer/modeling_longformer.py +++ b/src/transformers/models/longformer/modeling_longformer.py @@ -1539,8 +1539,7 @@ def forward( token_type_ids: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerBaseModelOutputWithPooling]: r""" @@ -1691,8 +1690,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerMaskedLMOutput]: r""" @@ -1804,8 +1802,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerSequenceClassifierOutput]: r""" @@ -1928,8 +1925,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerQuestionAnsweringModelOutput]: r""" @@ -2058,8 +2054,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerTokenClassifierOutput]: r""" @@ -2139,8 +2134,7 @@ def forward( labels: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, LongformerMultipleChoiceModelOutput]: r""" diff --git a/src/transformers/models/marian/modeling_marian.py b/src/transformers/models/marian/modeling_marian.py index 5630f916ee38..48390d047539 100755 --- a/src/transformers/models/marian/modeling_marian.py +++ b/src/transformers/models/marian/modeling_marian.py @@ -739,8 +739,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutput]: r""" @@ -910,8 +909,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: @@ -1250,8 +1248,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Seq2SeqModelOutput: @@ -1508,8 +1505,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Seq2SeqLMOutput: @@ -1688,8 +1684,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/mistral/modular_mistral.py b/src/transformers/models/mistral/modular_mistral.py index 0e7238fdef52..7aa6b0b60efe 100644 --- a/src/transformers/models/mistral/modular_mistral.py +++ b/src/transformers/models/mistral/modular_mistral.py @@ -116,8 +116,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -243,8 +242,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: r""" diff --git a/src/transformers/models/mistral3/modular_mistral3.py b/src/transformers/models/mistral3/modular_mistral3.py index 391b02d42b8f..09cb5808712e 100644 --- a/src/transformers/models/mistral3/modular_mistral3.py +++ b/src/transformers/models/mistral3/modular_mistral3.py @@ -185,8 +185,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, vision_feature_layer: Optional[Union[int, list[int]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, image_sizes: torch.Tensor = None, @@ -278,8 +277,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/mlcd/modular_mlcd.py b/src/transformers/models/mlcd/modular_mlcd.py index d18b2346224a..600324532a30 100644 --- a/src/transformers/models/mlcd/modular_mlcd.py +++ b/src/transformers/models/mlcd/modular_mlcd.py @@ -315,8 +315,7 @@ def forward( inputs_embeds: torch.FloatTensor, position_embeddings: tuple[torch.Tensor, torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -399,8 +398,7 @@ def __init__(self, config: MLCDVisionConfig): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_hidden_states = ( @@ -491,8 +489,7 @@ class MLCDVisionModel(CLIPVisionModel): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" diff --git a/src/transformers/models/mllama/modeling_mllama.py b/src/transformers/models/mllama/modeling_mllama.py index 841674f0c7fb..91c8ffac8231 100644 --- a/src/transformers/models/mllama/modeling_mllama.py +++ b/src/transformers/models/mllama/modeling_mllama.py @@ -349,8 +349,7 @@ def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1084,8 +1083,7 @@ def forward( pixel_values: torch.Tensor, aspect_ratio_ids: torch.Tensor, aspect_ratio_mask: torch.Tensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BaseModelOutput, tuple[torch.Tensor, ...]]: r""" @@ -1298,8 +1296,7 @@ def forward( past_key_values: Optional[Union[Cache, list[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1513,8 +1510,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, @@ -1656,8 +1652,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1822,8 +1817,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/modernbert/modular_modernbert.py b/src/transformers/models/modernbert/modular_modernbert.py index ff46a523a6cc..72354be49248 100644 --- a/src/transformers/models/modernbert/modular_modernbert.py +++ b/src/transformers/models/modernbert/modular_modernbert.py @@ -920,8 +920,7 @@ def forward( max_seqlen: Optional[int] = None, batch_size: Optional[int] = None, seq_len: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor, ...], BaseModelOutput]: r""" @@ -1131,8 +1130,7 @@ def forward( max_seqlen: Optional[int] = None, batch_size: Optional[int] = None, seq_len: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: @@ -1263,8 +1261,7 @@ def forward( max_seqlen: Optional[int] = None, batch_size: Optional[int] = None, seq_len: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: @@ -1386,8 +1383,7 @@ def forward( max_seqlen: Optional[int] = None, batch_size: Optional[int] = None, seq_len: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1476,8 +1472,7 @@ def forward( max_seqlen: Optional[int] = None, batch_size: Optional[int] = None, seq_len: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: diff --git a/src/transformers/models/moonshine/modular_moonshine.py b/src/transformers/models/moonshine/modular_moonshine.py index ea82ce3af3b3..8faca06f6274 100644 --- a/src/transformers/models/moonshine/modular_moonshine.py +++ b/src/transformers/models/moonshine/modular_moonshine.py @@ -578,8 +578,7 @@ def forward( self, input_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" @@ -694,8 +693,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, @@ -836,8 +834,7 @@ def forward( decoder_inputs_embeds: Optional[tuple[torch.FloatTensor]] = None, decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, ) -> Seq2SeqModelOutput: r""" @@ -995,8 +992,7 @@ def forward( decoder_inputs_embeds: Optional[tuple[torch.FloatTensor]] = None, decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, ) -> Seq2SeqLMOutput: diff --git a/src/transformers/models/mpt/modeling_mpt.py b/src/transformers/models/mpt/modeling_mpt.py index 79ec42e2b8d8..bb2e89457a75 100644 --- a/src/transformers/models/mpt/modeling_mpt.py +++ b/src/transformers/models/mpt/modeling_mpt.py @@ -313,8 +313,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple[torch.Tensor, ...], BaseModelOutputWithPastAndCrossAttentions]: @@ -464,8 +463,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -583,8 +581,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutputWithPast]: r""" @@ -706,8 +703,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **deprecated_arguments, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: @@ -785,8 +781,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/mt5/modeling_mt5.py b/src/transformers/models/mt5/modeling_mt5.py index 8596fbeb4f9d..d2c7f5430c98 100644 --- a/src/transformers/models/mt5/modeling_mt5.py +++ b/src/transformers/models/mt5/modeling_mt5.py @@ -1441,8 +1441,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: @@ -1714,8 +1713,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: @@ -2025,8 +2023,7 @@ def forward( attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutput]: r""" @@ -2104,8 +2101,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: r""" @@ -2262,8 +2258,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -2380,8 +2375,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqQuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/musicgen/modeling_musicgen.py b/src/transformers/models/musicgen/modeling_musicgen.py index a7ead0a51461..0e9781aa5145 100644 --- a/src/transformers/models/musicgen/modeling_musicgen.py +++ b/src/transformers/models/musicgen/modeling_musicgen.py @@ -500,8 +500,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -782,8 +781,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -908,8 +906,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1713,8 +1710,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Seq2SeqLMOutput]: diff --git a/src/transformers/models/mvp/modeling_mvp.py b/src/transformers/models/mvp/modeling_mvp.py index 739ecd8f015e..c1db3afe7d57 100644 --- a/src/transformers/models/mvp/modeling_mvp.py +++ b/src/transformers/models/mvp/modeling_mvp.py @@ -580,8 +580,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -782,8 +781,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -1052,8 +1050,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqModelOutput]: r""" @@ -1220,8 +1217,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqLMOutput]: r""" @@ -1392,8 +1388,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: r""" @@ -1565,8 +1560,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqQuestionAnsweringModelOutput]: r""" @@ -1763,8 +1757,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: r""" diff --git a/src/transformers/models/nemotron/modeling_nemotron.py b/src/transformers/models/nemotron/modeling_nemotron.py index e1c1c82e0b47..8c436fa83863 100644 --- a/src/transformers/models/nemotron/modeling_nemotron.py +++ b/src/transformers/models/nemotron/modeling_nemotron.py @@ -637,8 +637,7 @@ def forward( past_key_values: Optional[Union[Cache, list[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -907,8 +906,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **loss_kwargs, @@ -1013,8 +1011,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -1104,8 +1101,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.transformer( @@ -1174,8 +1170,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/nllb_moe/modeling_nllb_moe.py b/src/transformers/models/nllb_moe/modeling_nllb_moe.py index 15c87871649b..1f6ef1462c01 100644 --- a/src/transformers/models/nllb_moe/modeling_nllb_moe.py +++ b/src/transformers/models/nllb_moe/modeling_nllb_moe.py @@ -922,8 +922,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ): @@ -1139,8 +1138,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ): @@ -1499,8 +1497,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqMoEModelOutput]: @@ -1648,8 +1645,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqMoEOutput]: diff --git a/src/transformers/models/olmoe/modeling_olmoe.py b/src/transformers/models/olmoe/modeling_olmoe.py index 6dc3c12c1ffb..dbb81cddc219 100644 --- a/src/transformers/models/olmoe/modeling_olmoe.py +++ b/src/transformers/models/olmoe/modeling_olmoe.py @@ -757,8 +757,7 @@ def forward( past_key_values: Optional[Union[Cache, list[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1049,8 +1048,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py b/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py index 77af9993033f..c56c3415426e 100644 --- a/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py +++ b/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py @@ -1520,8 +1520,7 @@ def forward( tasks_attention_mask: torch.LongTensor, classes_structure: torch.LongTensor, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], OmDetTurboObjectDetectionOutput]: r""" diff --git a/src/transformers/models/openai/modeling_openai.py b/src/transformers/models/openai/modeling_openai.py index 73aba2143713..e57d9f8a8964 100644 --- a/src/transformers/models/openai/modeling_openai.py +++ b/src/transformers/models/openai/modeling_openai.py @@ -450,8 +450,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -567,8 +566,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutput]: @@ -660,8 +658,7 @@ def forward( mc_token_ids: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, mc_labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], OpenAIGPTDoubleHeadsModelOutput]: r""" @@ -772,8 +769,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/opt/modeling_opt.py b/src/transformers/models/opt/modeling_opt.py index fd22722b69cf..0dd16645b5b4 100644 --- a/src/transformers/models/opt/modeling_opt.py +++ b/src/transformers/models/opt/modeling_opt.py @@ -522,8 +522,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, cache_position: Optional[torch.Tensor] = None, @@ -754,8 +753,7 @@ def forward( past_key_values: Optional[Union[list[torch.FloatTensor], Cache]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, cache_position: Optional[torch.Tensor] = None, @@ -837,8 +835,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, cache_position: Optional[torch.Tensor] = None, @@ -954,8 +951,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: @@ -1067,8 +1063,7 @@ def forward( start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: diff --git a/src/transformers/models/owlv2/modeling_owlv2.py b/src/transformers/models/owlv2/modeling_owlv2.py index 4e838c90a9f8..5d8065fdae0a 100644 --- a/src/transformers/models/owlv2/modeling_owlv2.py +++ b/src/transformers/models/owlv2/modeling_owlv2.py @@ -616,8 +616,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -702,8 +701,7 @@ def forward( input_ids: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -784,8 +782,7 @@ def forward( self, input_ids: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -833,8 +830,7 @@ def __init__(self, config: Owlv2VisionConfig): def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -892,8 +888,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -966,8 +961,7 @@ def get_text_features( self, input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1005,8 +999,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> torch.FloatTensor: @@ -1055,8 +1048,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_base_image_embeds: Optional[bool] = None, return_dict: Optional[bool] = None, @@ -1345,8 +1337,7 @@ def image_text_embedder( input_ids: torch.Tensor, pixel_values: torch.FloatTensor, attention_mask: torch.Tensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> tuple[torch.FloatTensor]: # Encode text and image @@ -1395,8 +1386,7 @@ def image_text_embedder( def image_embedder( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> tuple[torch.FloatTensor]: # Get Owlv2Model vision embeddings (same as CLIP) @@ -1484,8 +1474,7 @@ def image_guided_detection( self, pixel_values: torch.FloatTensor, query_pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Owlv2ImageGuidedObjectDetectionOutput: @@ -1603,8 +1592,7 @@ def forward( input_ids: torch.Tensor, pixel_values: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Owlv2ObjectDetectionOutput: diff --git a/src/transformers/models/owlvit/modeling_owlvit.py b/src/transformers/models/owlvit/modeling_owlvit.py index d487706611d1..ff3b8a5b0424 100644 --- a/src/transformers/models/owlvit/modeling_owlvit.py +++ b/src/transformers/models/owlvit/modeling_owlvit.py @@ -602,8 +602,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -687,8 +686,7 @@ def forward( input_ids: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -768,8 +766,7 @@ def forward( self, input_ids: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -816,8 +813,7 @@ def __init__(self, config: OwlViTVisionConfig): def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -874,8 +870,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -947,8 +942,7 @@ def get_text_features( self, input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -986,8 +980,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> torch.FloatTensor: @@ -1036,8 +1029,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_base_image_embeds: Optional[bool] = None, return_dict: Optional[bool] = None, @@ -1304,8 +1296,7 @@ def image_text_embedder( input_ids: torch.Tensor, pixel_values: torch.FloatTensor, attention_mask: torch.Tensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> tuple[torch.FloatTensor]: # Encode text and image @@ -1353,8 +1344,7 @@ def image_text_embedder( def image_embedder( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> tuple[torch.FloatTensor]: # Get OwlViTModel vision embeddings (same as CLIP) @@ -1441,8 +1431,7 @@ def image_guided_detection( self, pixel_values: torch.FloatTensor, query_pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> OwlViTImageGuidedObjectDetectionOutput: @@ -1545,8 +1534,7 @@ def forward( input_ids: torch.Tensor, pixel_values: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> OwlViTObjectDetectionOutput: diff --git a/src/transformers/models/paligemma/modeling_paligemma.py b/src/transformers/models/paligemma/modeling_paligemma.py index 64055cbe1c7e..4fd59aca1f5b 100644 --- a/src/transformers/models/paligemma/modeling_paligemma.py +++ b/src/transformers/models/paligemma/modeling_paligemma.py @@ -271,8 +271,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, PaligemmaModelOutputWithPast]: @@ -453,8 +452,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], diff --git a/src/transformers/models/pegasus/modeling_pegasus.py b/src/transformers/models/pegasus/modeling_pegasus.py index 6922d3c815bf..b8e69f957aa6 100755 --- a/src/transformers/models/pegasus/modeling_pegasus.py +++ b/src/transformers/models/pegasus/modeling_pegasus.py @@ -1268,8 +1268,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, Seq2SeqModelOutput]: @@ -1455,8 +1454,7 @@ def forward( decoder_inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: @@ -1658,8 +1656,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/pegasus_x/modeling_pegasus_x.py b/src/transformers/models/pegasus_x/modeling_pegasus_x.py index 0c48aa614125..90fe66934711 100755 --- a/src/transformers/models/pegasus_x/modeling_pegasus_x.py +++ b/src/transformers/models/pegasus_x/modeling_pegasus_x.py @@ -1516,8 +1516,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, Seq2SeqModelOutput]: @@ -1673,8 +1672,7 @@ def forward( decoder_inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: diff --git a/src/transformers/models/perceiver/modeling_perceiver.py b/src/transformers/models/perceiver/modeling_perceiver.py index 1a6f65cf6c44..c700f0a7ae14 100755 --- a/src/transformers/models/perceiver/modeling_perceiver.py +++ b/src/transformers/models/perceiver/modeling_perceiver.py @@ -686,8 +686,7 @@ def forward( attention_mask: Optional[torch.FloatTensor] = None, subsampled_output_points: Optional[dict[str, torch.Tensor]] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, PerceiverModelOutput]: @@ -929,8 +928,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, input_ids: Optional[torch.Tensor] = None, @@ -1056,8 +1054,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, input_ids: Optional[torch.Tensor] = None, @@ -1189,8 +1186,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, @@ -1333,8 +1329,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, @@ -1475,8 +1470,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, @@ -1634,8 +1628,7 @@ def forward( inputs: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, PerceiverClassifierOutput]: @@ -1844,8 +1837,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, subsampled_output_points: Optional[dict[str, torch.Tensor]] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, PerceiverClassifierOutput]: diff --git a/src/transformers/models/persimmon/modeling_persimmon.py b/src/transformers/models/persimmon/modeling_persimmon.py index 3b29d8e4e753..411c0612d90f 100644 --- a/src/transformers/models/persimmon/modeling_persimmon.py +++ b/src/transformers/models/persimmon/modeling_persimmon.py @@ -453,8 +453,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -741,8 +740,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, @@ -854,8 +852,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -950,8 +947,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/phi/modular_phi.py b/src/transformers/models/phi/modular_phi.py index 762318773664..0aef770b1e40 100644 --- a/src/transformers/models/phi/modular_phi.py +++ b/src/transformers/models/phi/modular_phi.py @@ -204,8 +204,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index b708bcb999b5..31568b1aad8f 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -407,8 +407,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -678,8 +677,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -773,8 +771,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py b/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py index 28cd2de9fbca..965cbb99a4b2 100644 --- a/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py +++ b/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py @@ -670,8 +670,7 @@ def forward( self, pixel_values, patch_attention_mask: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -1509,8 +1508,7 @@ def forward( audio_embed_sizes=None, audio_attention_mask=None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs, ) -> BaseModelOutputWithPast: @@ -1650,8 +1648,7 @@ def forward( audio_attention_mask=None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, diff --git a/src/transformers/models/phimoe/modeling_phimoe.py b/src/transformers/models/phimoe/modeling_phimoe.py index f4d709343307..2fe0bb916e04 100644 --- a/src/transformers/models/phimoe/modeling_phimoe.py +++ b/src/transformers/models/phimoe/modeling_phimoe.py @@ -944,8 +944,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: @@ -1273,8 +1272,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, @@ -1441,8 +1439,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" diff --git a/src/transformers/models/pix2struct/modeling_pix2struct.py b/src/transformers/models/pix2struct/modeling_pix2struct.py index 54c956601a51..63f07293d0f1 100644 --- a/src/transformers/models/pix2struct/modeling_pix2struct.py +++ b/src/transformers/models/pix2struct/modeling_pix2struct.py @@ -519,8 +519,7 @@ def forward( flattened_patches: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1105,8 +1104,7 @@ def forward( cross_attn_head_mask: Optional[torch.Tensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.LongTensor] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1525,8 +1523,7 @@ def forward( labels: Optional[torch.LongTensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: diff --git a/src/transformers/models/plbart/modular_plbart.py b/src/transformers/models/plbart/modular_plbart.py index e5ec69efcf05..5881ab8ae902 100644 --- a/src/transformers/models/plbart/modular_plbart.py +++ b/src/transformers/models/plbart/modular_plbart.py @@ -328,8 +328,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: @@ -486,8 +485,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqLMOutput]: diff --git a/src/transformers/models/pop2piano/modeling_pop2piano.py b/src/transformers/models/pop2piano/modeling_pop2piano.py index b5e52e7c3971..778e802e9678 100644 --- a/src/transformers/models/pop2piano/modeling_pop2piano.py +++ b/src/transformers/models/pop2piano/modeling_pop2piano.py @@ -1161,8 +1161,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: diff --git a/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py b/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py index 9563fa98846c..e5437b85cf66 100644 --- a/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py +++ b/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py @@ -239,8 +239,7 @@ def forward( pixel_values: torch.FloatTensor, prompt_depth: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], DepthEstimatorOutput]: r""" diff --git a/src/transformers/models/prophetnet/modeling_prophetnet.py b/src/transformers/models/prophetnet/modeling_prophetnet.py index d7783c48e0aa..17a48cd91c28 100644 --- a/src/transformers/models/prophetnet/modeling_prophetnet.py +++ b/src/transformers/models/prophetnet/modeling_prophetnet.py @@ -1124,8 +1124,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1267,8 +1266,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ProphetNetDecoderModelOutput]: r""" @@ -1615,8 +1613,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ProphetNetSeq2SeqModelOutput]: r""" @@ -1756,8 +1753,7 @@ def forward( decoder_inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ProphetNetSeq2SeqLMOutput]: r""" @@ -1975,8 +1971,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ProphetNetDecoderLMOutput]: r""" diff --git a/src/transformers/models/pvt_v2/modeling_pvt_v2.py b/src/transformers/models/pvt_v2/modeling_pvt_v2.py index e378ce390abc..33d3ccb7e931 100644 --- a/src/transformers/models/pvt_v2/modeling_pvt_v2.py +++ b/src/transformers/models/pvt_v2/modeling_pvt_v2.py @@ -438,8 +438,7 @@ class PreTrainedModel def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -492,8 +491,7 @@ def forward( self, pixel_values: Optional[torch.Tensor], labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: r""" @@ -574,8 +572,7 @@ def __init__(self, config: PvtV2Config): def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> BackboneOutput: r""" diff --git a/src/transformers/models/qwen2/modular_qwen2.py b/src/transformers/models/qwen2/modular_qwen2.py index 90e22f4507c1..ee80e6a24582 100644 --- a/src/transformers/models/qwen2/modular_qwen2.py +++ b/src/transformers/models/qwen2/modular_qwen2.py @@ -119,8 +119,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: diff --git a/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py b/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py index 58b3b87ad381..78ae61168d8f 100644 --- a/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py +++ b/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py @@ -2356,8 +2356,7 @@ def forward( rope_deltas: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, use_audio_in_video: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -2695,8 +2694,7 @@ def forward( use_audio_in_video: Optional[bool] = None, audio_feature_lengths: Optional[torch.LongTensor] = None, video_second_per_grid: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Qwen2_5OmniTalkerCausalLMOutputWithPast]: r""" diff --git a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py index 71764b245633..cf8e391cc03a 100644 --- a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py +++ b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py @@ -613,8 +613,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, pixel_values_videos: Optional[torch.FloatTensor] = None, diff --git a/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py b/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py index 6569d78674e9..954d74081235 100644 --- a/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py +++ b/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py @@ -749,8 +749,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Qwen2AudioCausalLMOutputWithPast]: r""" diff --git a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py index 00db18aaf37b..ff0fb2dd526e 100644 --- a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py +++ b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py @@ -792,8 +792,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: @@ -1119,8 +1118,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, @@ -1244,8 +1242,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -1340,8 +1337,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" @@ -1407,8 +1403,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: r""" diff --git a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py index 95687334eb0f..a0ac1b0098ce 100644 --- a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py +++ b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py @@ -884,8 +884,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1225,8 +1224,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, pixel_values_videos: Optional[torch.FloatTensor] = None, diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 27190d6ba524..5caeeab64b0e 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -378,8 +378,7 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: @@ -622,8 +621,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -717,8 +715,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" @@ -783,8 +780,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.transformer( diff --git a/src/transformers/models/qwen3_moe/modular_qwen3_moe.py b/src/transformers/models/qwen3_moe/modular_qwen3_moe.py index 569b92cc6e85..19780363f229 100644 --- a/src/transformers/models/qwen3_moe/modular_qwen3_moe.py +++ b/src/transformers/models/qwen3_moe/modular_qwen3_moe.py @@ -242,8 +242,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/rag/modeling_rag.py b/src/transformers/models/rag/modeling_rag.py index 5729e49d310d..15d9fc61e882 100644 --- a/src/transformers/models/rag/modeling_rag.py +++ b/src/transformers/models/rag/modeling_rag.py @@ -440,8 +440,7 @@ def forward( context_input_ids: Optional[torch.LongTensor] = None, context_attention_mask: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_retrieved: Optional[bool] = None, n_docs: Optional[int] = None, ) -> Union[tuple[torch.Tensor], RetrievAugLMOutput]: @@ -714,8 +713,7 @@ def forward( context_attention_mask: Optional[torch.LongTensor] = None, doc_scores: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_retrieved: Optional[bool] = None, exclude_bos_score: Optional[bool] = None, reduce_loss: Optional[bool] = None, @@ -1224,8 +1222,7 @@ def forward( context_attention_mask: Optional[torch.LongTensor] = None, doc_scores: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_retrieved: Optional[bool] = None, do_marginalize: Optional[bool] = None, reduce_loss: Optional[bool] = None, diff --git a/src/transformers/models/rembert/modeling_rembert.py b/src/transformers/models/rembert/modeling_rembert.py index 22a3ba7aeb96..f6140b7a062e 100755 --- a/src/transformers/models/rembert/modeling_rembert.py +++ b/src/transformers/models/rembert/modeling_rembert.py @@ -707,8 +707,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -837,8 +836,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -945,8 +943,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1052,8 +1049,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1137,8 +1133,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1242,8 +1237,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1310,8 +1304,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/roberta/modeling_roberta.py b/src/transformers/models/roberta/modeling_roberta.py index 9381c8f9ab01..39ce9d2e38b8 100644 --- a/src/transformers/models/roberta/modeling_roberta.py +++ b/src/transformers/models/roberta/modeling_roberta.py @@ -763,8 +763,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -934,8 +933,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1063,8 +1061,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1178,8 +1175,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1272,8 +1268,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1382,8 +1377,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1482,8 +1476,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py b/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py index 31d459e7d8b6..2ca11d1b15fe 100644 --- a/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py +++ b/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py @@ -648,8 +648,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -802,8 +801,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -937,8 +935,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1054,8 +1051,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1149,8 +1145,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1260,8 +1255,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1362,8 +1356,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/roc_bert/modeling_roc_bert.py b/src/transformers/models/roc_bert/modeling_roc_bert.py index 06721ae7719b..d1e199d5d4b2 100644 --- a/src/transformers/models/roc_bert/modeling_roc_bert.py +++ b/src/transformers/models/roc_bert/modeling_roc_bert.py @@ -838,8 +838,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -1002,8 +1001,7 @@ def forward( labels_input_pronunciation_ids: Optional[torch.Tensor] = None, labels_attention_mask: Optional[torch.Tensor] = None, labels_token_type_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: @@ -1208,8 +1206,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1364,8 +1361,7 @@ def forward( past_key_values: Optional[list[torch.Tensor]] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1538,8 +1534,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1645,8 +1640,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1778,8 +1772,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1864,8 +1857,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/roformer/modeling_roformer.py b/src/transformers/models/roformer/modeling_roformer.py index 3f9b2875c207..73afcf6858f7 100644 --- a/src/transformers/models/roformer/modeling_roformer.py +++ b/src/transformers/models/roformer/modeling_roformer.py @@ -846,8 +846,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPastAndCrossAttentions, tuple[torch.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -973,8 +972,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MaskedLMOutput, tuple[torch.Tensor]]: r""" @@ -1073,8 +1071,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[CausalLMOutputWithCrossAttentions, tuple[torch.Tensor]]: @@ -1198,8 +1195,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.Tensor]]: r""" @@ -1279,8 +1275,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MultipleChoiceModelOutput, tuple[torch.Tensor]]: r""" @@ -1377,8 +1372,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[TokenClassifierOutput, tuple[torch.Tensor]]: r""" @@ -1444,8 +1438,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[QuestionAnsweringModelOutput, tuple[torch.Tensor]]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/rwkv/modeling_rwkv.py b/src/transformers/models/rwkv/modeling_rwkv.py index b7362000b572..edcea72ab8ab 100644 --- a/src/transformers/models/rwkv/modeling_rwkv.py +++ b/src/transformers/models/rwkv/modeling_rwkv.py @@ -538,8 +538,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, state: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, RwkvOutput]: r""" @@ -736,8 +735,7 @@ def forward( state: Optional[list[torch.FloatTensor]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, RwkvCausalLMOutput]: diff --git a/src/transformers/models/sam/modeling_sam.py b/src/transformers/models/sam/modeling_sam.py index 28e28f07d2d5..733854f43ef1 100644 --- a/src/transformers/models/sam/modeling_sam.py +++ b/src/transformers/models/sam/modeling_sam.py @@ -1123,8 +1123,7 @@ def get_input_embeddings(self): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> SamVisionEncoderOutput: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -1221,8 +1220,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SamVisionEncoderOutput]: return self.vision_encoder( @@ -1278,8 +1276,7 @@ def get_image_wide_positional_embeddings(self): def get_image_embeddings( self, pixel_values, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ): r""" Returns the image embeddings by passing the pixel values through the vision encoder. @@ -1346,8 +1343,7 @@ def forward( multimask_output: bool = True, attention_similarity: Optional[torch.FloatTensor] = None, target_embedding: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + **kwargs, ) -> SamImageSegmentationOutput: r""" diff --git a/src/transformers/models/sam_hq/modular_sam_hq.py b/src/transformers/models/sam_hq/modular_sam_hq.py index 45d8067d83af..62b927059d6f 100644 --- a/src/transformers/models/sam_hq/modular_sam_hq.py +++ b/src/transformers/models/sam_hq/modular_sam_hq.py @@ -130,8 +130,7 @@ class SamHQVisionEncoder(SamVisionEncoder): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SamHQVisionEncoderOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -451,8 +450,7 @@ def __init__(self, config): def get_image_embeddings( self, pixel_values, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): r""" @@ -492,8 +490,7 @@ def forward( hq_token_only: bool = False, attention_similarity: Optional[torch.FloatTensor] = None, target_embedding: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, intermediate_embeddings: Optional[list[torch.FloatTensor]] = None, **kwargs, diff --git a/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py b/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py index 03136936e60c..4addfe42e0fc 100755 --- a/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py +++ b/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py @@ -1485,8 +1485,7 @@ def forward( self, input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Wav2Vec2BaseModelOutput]: @@ -1600,8 +1599,7 @@ def forward( input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, BaseModelOutput]: @@ -1794,8 +1792,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1954,8 +1951,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -2079,8 +2075,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: r""" @@ -2564,8 +2559,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -2832,8 +2826,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -3117,8 +3110,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: r""" @@ -3449,8 +3441,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -3819,8 +3810,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: diff --git a/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py b/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py index b6047a10da60..8891bd88777f 100644 --- a/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py +++ b/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py @@ -1553,8 +1553,7 @@ def forward( self, input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Wav2Vec2BaseModelOutput]: @@ -1669,8 +1668,7 @@ def forward( input_ids: Optional[torch.LongTensor] = None, attention_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, BaseModelOutput]: @@ -1864,8 +1862,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -2069,8 +2066,7 @@ def forward( char_input_ids: Optional[torch.LongTensor] = None, char_count_per_id: Optional[torch.LongTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SeamlessM4Tv2TextToUnitDecoderOutput]: r""" @@ -2205,8 +2201,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -2328,8 +2323,7 @@ def forward( encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -2809,8 +2803,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -3086,8 +3079,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -3382,8 +3374,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: r""" @@ -3755,8 +3746,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: @@ -4165,8 +4155,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[Seq2SeqLMOutput, tuple[torch.FloatTensor]]: diff --git a/src/transformers/models/segformer/modeling_segformer.py b/src/transformers/models/segformer/modeling_segformer.py index 959601aa57a0..5732a06b72a2 100755 --- a/src/transformers/models/segformer/modeling_segformer.py +++ b/src/transformers/models/segformer/modeling_segformer.py @@ -472,8 +472,7 @@ class PreTrainedModel def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -524,8 +523,7 @@ def forward( self, pixel_values: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SegFormerImageClassifierOutput]: r""" @@ -683,8 +681,7 @@ def forward( self, pixel_values: torch.FloatTensor, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SemanticSegmenterOutput]: r""" diff --git a/src/transformers/models/seggpt/modeling_seggpt.py b/src/transformers/models/seggpt/modeling_seggpt.py index 9a7d830476b5..855096e12717 100644 --- a/src/transformers/models/seggpt/modeling_seggpt.py +++ b/src/transformers/models/seggpt/modeling_seggpt.py @@ -681,8 +681,7 @@ def forward( feature_ensemble: Optional[bool] = None, embedding_type: Optional[str] = None, labels: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SegGptEncoderOutput]: r""" @@ -877,8 +876,7 @@ def forward( feature_ensemble: Optional[bool] = None, embedding_type: Optional[str] = None, labels: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SegGptImageSegmentationOutput]: r""" diff --git a/src/transformers/models/sew/modular_sew.py b/src/transformers/models/sew/modular_sew.py index 0b151b05a2ce..bb85bb1ea18a 100644 --- a/src/transformers/models/sew/modular_sew.py +++ b/src/transformers/models/sew/modular_sew.py @@ -409,8 +409,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" diff --git a/src/transformers/models/sew_d/modeling_sew_d.py b/src/transformers/models/sew_d/modeling_sew_d.py index f2d682884c47..9bf50bb90e40 100644 --- a/src/transformers/models/sew_d/modeling_sew_d.py +++ b/src/transformers/models/sew_d/modeling_sew_d.py @@ -1343,8 +1343,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1481,8 +1480,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -1605,8 +1603,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: diff --git a/src/transformers/models/shieldgemma2/modeling_shieldgemma2.py b/src/transformers/models/shieldgemma2/modeling_shieldgemma2.py index a164e61420a6..958beef19e47 100644 --- a/src/transformers/models/shieldgemma2/modeling_shieldgemma2.py +++ b/src/transformers/models/shieldgemma2/modeling_shieldgemma2.py @@ -86,8 +86,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **lm_kwargs, diff --git a/src/transformers/models/siglip/modeling_siglip.py b/src/transformers/models/siglip/modeling_siglip.py index 55a06643b72d..6ca0c988860b 100644 --- a/src/transformers/models/siglip/modeling_siglip.py +++ b/src/transformers/models/siglip/modeling_siglip.py @@ -589,8 +589,7 @@ def forward( self, inputs_embeds, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutput: r""" Args: @@ -667,8 +666,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -738,8 +736,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: r""" Examples: @@ -785,8 +782,7 @@ def __init__(self, config: SiglipVisionConfig): def forward( self, pixel_values, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = False, ) -> BaseModelOutputWithPooling: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -864,8 +860,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> BaseModelOutputWithPooling: r""" @@ -939,8 +934,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> torch.FloatTensor: r""" Returns: @@ -983,8 +977,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> torch.FloatTensor: r""" @@ -1037,8 +1030,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> SiglipOutput: r""" @@ -1160,8 +1152,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, ) -> ImageClassifierOutput: r""" diff --git a/src/transformers/models/siglip2/modular_siglip2.py b/src/transformers/models/siglip2/modular_siglip2.py index 803c1b070b3f..4361fff5c285 100644 --- a/src/transformers/models/siglip2/modular_siglip2.py +++ b/src/transformers/models/siglip2/modular_siglip2.py @@ -241,8 +241,7 @@ def forward( pixel_values: torch.FloatTensor, attention_mask: torch.Tensor, spatial_shapes: torch.LongTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: r""" spatial_shapes (`torch.LongTensor` of shape `(batch_size, 2)`): @@ -320,8 +319,7 @@ def forward( pixel_values: torch.FloatTensor, pixel_attention_mask: torch.Tensor, spatial_shapes: torch.LongTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> BaseModelOutputWithPooling: r""" pixel_attention_mask (`torch.Tensor` of shape `(batch_size, image_size, image_size)`, *optional*): @@ -364,8 +362,7 @@ def get_image_features( pixel_values: Optional[torch.FloatTensor] = None, pixel_attention_mask: Optional[torch.Tensor] = None, spatial_shapes: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> torch.FloatTensor: r""" pixel_attention_mask (`torch.Tensor` of shape `(batch_size, image_size, image_size)`, *optional*): @@ -425,8 +422,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> Siglip2Output: r""" pixel_attention_mask (`torch.Tensor` of shape `(batch_size, image_size, image_size)`, *optional*): @@ -528,8 +524,7 @@ def forward( pixel_attention_mask: Optional[torch.Tensor] = None, spatial_shapes: Optional[torch.LongTensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> ImageClassifierOutput: r""" pixel_attention_mask (`torch.Tensor` of shape `(batch_size, image_size, image_size)`, *optional*): diff --git a/src/transformers/models/smolvlm/modular_smolvlm.py b/src/transformers/models/smolvlm/modular_smolvlm.py index 32fb122563f7..6d9d19183825 100644 --- a/src/transformers/models/smolvlm/modular_smolvlm.py +++ b/src/transformers/models/smolvlm/modular_smolvlm.py @@ -264,8 +264,7 @@ def forward( pixel_attention_mask: Optional[torch.BoolTensor] = None, image_hidden_states: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], diff --git a/src/transformers/models/speech_encoder_decoder/modeling_speech_encoder_decoder.py b/src/transformers/models/speech_encoder_decoder/modeling_speech_encoder_decoder.py index b4582949fe6c..da09cdd978e6 100644 --- a/src/transformers/models/speech_encoder_decoder/modeling_speech_encoder_decoder.py +++ b/src/transformers/models/speech_encoder_decoder/modeling_speech_encoder_decoder.py @@ -329,8 +329,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + input_values: Optional[torch.FloatTensor] = None, input_features: Optional[torch.FloatTensor] = None, return_dict: Optional[bool] = None, diff --git a/src/transformers/models/speech_to_text/modeling_speech_to_text.py b/src/transformers/models/speech_to_text/modeling_speech_to_text.py index d9fbf2faec91..8a44f497b56a 100755 --- a/src/transformers/models/speech_to_text/modeling_speech_to_text.py +++ b/src/transformers/models/speech_to_text/modeling_speech_to_text.py @@ -1113,8 +1113,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: r""" @@ -1275,8 +1274,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: r""" diff --git a/src/transformers/models/speecht5/modeling_speecht5.py b/src/transformers/models/speecht5/modeling_speecht5.py index 24b833682159..fd4316775b7d 100644 --- a/src/transformers/models/speecht5/modeling_speecht5.py +++ b/src/transformers/models/speecht5/modeling_speecht5.py @@ -1266,8 +1266,7 @@ def forward( hidden_states: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: """ @@ -1395,8 +1394,7 @@ def forward( input_values: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: hidden_states, attention_mask = self.prenet(input_values, attention_mask) @@ -1437,8 +1435,7 @@ def forward( input_values: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: hidden_states = self.prenet(input_values) @@ -1473,8 +1470,7 @@ def forward( input_values: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: return self.wrapped_encoder( @@ -1513,8 +1509,7 @@ def forward( cross_attn_head_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -1719,8 +1714,7 @@ def forward( cross_attn_head_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: decoder_hidden_states = self.prenet(input_values, speaker_embeddings) @@ -1771,8 +1765,7 @@ def forward( cross_attn_head_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: decoder_hidden_states, attention_mask = self.prenet(input_values, attention_mask, past_key_values) @@ -1817,8 +1810,7 @@ def forward( cross_attn_head_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: outputs = self.wrapped_decoder( @@ -2024,8 +2016,7 @@ def forward( past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, speaker_embeddings: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: r""" @@ -2179,8 +2170,7 @@ def forward( encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: @@ -2509,8 +2499,7 @@ def forward( encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, speaker_embeddings: Optional[torch.FloatTensor] = None, labels: Optional[torch.FloatTensor] = None, @@ -2862,8 +2851,7 @@ def forward( encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, speaker_embeddings: Optional[torch.FloatTensor] = None, labels: Optional[torch.FloatTensor] = None, diff --git a/src/transformers/models/splinter/modeling_splinter.py b/src/transformers/models/splinter/modeling_splinter.py index 06d2917b6a27..d1b09718e66b 100755 --- a/src/transformers/models/splinter/modeling_splinter.py +++ b/src/transformers/models/splinter/modeling_splinter.py @@ -578,8 +578,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -763,8 +762,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, question_positions: Optional[torch.LongTensor] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: @@ -916,8 +914,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, question_positions: Optional[torch.LongTensor] = None, ) -> Union[tuple, SplinterForPreTrainingOutput]: diff --git a/src/transformers/models/squeezebert/modeling_squeezebert.py b/src/transformers/models/squeezebert/modeling_squeezebert.py index 8175baf687b6..ef72dfb1e819 100644 --- a/src/transformers/models/squeezebert/modeling_squeezebert.py +++ b/src/transformers/models/squeezebert/modeling_squeezebert.py @@ -474,8 +474,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -564,8 +563,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -637,8 +635,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -723,8 +720,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -828,8 +824,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -895,8 +890,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/stablelm/modeling_stablelm.py b/src/transformers/models/stablelm/modeling_stablelm.py index a5399c90968a..14f503e41806 100755 --- a/src/transformers/models/stablelm/modeling_stablelm.py +++ b/src/transformers/models/stablelm/modeling_stablelm.py @@ -681,8 +681,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, ) -> BaseModelOutputWithPast: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -969,8 +968,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, @@ -1081,8 +1079,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -1177,8 +1174,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/models/starcoder2/modular_starcoder2.py b/src/transformers/models/starcoder2/modular_starcoder2.py index 517695595c54..fa92b519b8c3 100644 --- a/src/transformers/models/starcoder2/modular_starcoder2.py +++ b/src/transformers/models/starcoder2/modular_starcoder2.py @@ -172,8 +172,7 @@ def forward( past_key_values: Optional[Union[Cache, list[torch.FloatTensor]]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: diff --git a/src/transformers/models/superglue/modeling_superglue.py b/src/transformers/models/superglue/modeling_superglue.py index d597d04b649a..c34196cbd16c 100644 --- a/src/transformers/models/superglue/modeling_superglue.py +++ b/src/transformers/models/superglue/modeling_superglue.py @@ -749,8 +749,7 @@ def forward( self, pixel_values: torch.FloatTensor, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, KeypointMatchingOutput]: r""" diff --git a/src/transformers/models/swin/modeling_swin.py b/src/transformers/models/swin/modeling_swin.py index 99fa5b45dbcd..9713105de9c8 100644 --- a/src/transformers/models/swin/modeling_swin.py +++ b/src/transformers/models/swin/modeling_swin.py @@ -968,8 +968,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, SwinModelOutput]: @@ -1063,8 +1062,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, SwinMaskedImageModelingOutput]: @@ -1178,8 +1176,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, SwinImageClassifierOutput]: diff --git a/src/transformers/models/swin2sr/modeling_swin2sr.py b/src/transformers/models/swin2sr/modeling_swin2sr.py index 9a6030b4a0fb..4385dc59230c 100644 --- a/src/transformers/models/swin2sr/modeling_swin2sr.py +++ b/src/transformers/models/swin2sr/modeling_swin2sr.py @@ -808,8 +808,7 @@ def forward( self, pixel_values: torch.FloatTensor, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -1035,8 +1034,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageSuperResolutionOutput]: r""" diff --git a/src/transformers/models/swinv2/modeling_swinv2.py b/src/transformers/models/swinv2/modeling_swinv2.py index 0ad4821f0c31..9e6b401f7df8 100644 --- a/src/transformers/models/swinv2/modeling_swinv2.py +++ b/src/transformers/models/swinv2/modeling_swinv2.py @@ -1025,8 +1025,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, Swinv2ModelOutput]: @@ -1122,8 +1121,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, Swinv2MaskedImageModelingOutput]: @@ -1238,8 +1236,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, Swinv2ImageClassifierOutput]: @@ -1305,8 +1302,7 @@ def get_input_embeddings(self): def forward( self, pixel_values: Tensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> BackboneOutput: r""" diff --git a/src/transformers/models/switch_transformers/modeling_switch_transformers.py b/src/transformers/models/switch_transformers/modeling_switch_transformers.py index 833312311338..cd0e690304b8 100644 --- a/src/transformers/models/switch_transformers/modeling_switch_transformers.py +++ b/src/transformers/models/switch_transformers/modeling_switch_transformers.py @@ -1325,8 +1325,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = None, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1547,8 +1546,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = True, return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1846,8 +1844,7 @@ def forward( attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + output_router_logits: Optional[bool] = True, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], MoEModelOutput]: diff --git a/src/transformers/models/t5/modeling_t5.py b/src/transformers/models/t5/modeling_t5.py index 5339e8e2d904..057a52dd6995 100644 --- a/src/transformers/models/t5/modeling_t5.py +++ b/src/transformers/models/t5/modeling_t5.py @@ -1435,8 +1435,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: @@ -1687,8 +1686,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: @@ -1977,8 +1975,7 @@ def forward( attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutput]: r""" @@ -2054,8 +2051,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: r""" @@ -2210,8 +2206,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -2327,8 +2322,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqQuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/table_transformer/modeling_table_transformer.py b/src/transformers/models/table_transformer/modeling_table_transformer.py index 4938ea378dfb..51edf1fdefc0 100644 --- a/src/transformers/models/table_transformer/modeling_table_transformer.py +++ b/src/transformers/models/table_transformer/modeling_table_transformer.py @@ -1097,8 +1097,7 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], TableTransformerModelOutput]: r""" diff --git a/src/transformers/models/tapas/modeling_tapas.py b/src/transformers/models/tapas/modeling_tapas.py index f6463d95a24d..88e05f5f8fd3 100644 --- a/src/transformers/models/tapas/modeling_tapas.py +++ b/src/transformers/models/tapas/modeling_tapas.py @@ -775,8 +775,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -922,8 +921,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, MaskedLMOutput]: @@ -1063,8 +1061,7 @@ def forward( float_answer: Optional[torch.FloatTensor] = None, numeric_values: Optional[torch.FloatTensor] = None, numeric_values_scale: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TableQuestionAnsweringOutput]: r""" @@ -1386,8 +1383,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py b/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py index 83b69af4a10b..0f89eb8a6f24 100644 --- a/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py +++ b/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py @@ -784,8 +784,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -927,8 +926,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: @@ -1772,8 +1770,7 @@ def generate( past_observed_mask: Optional[torch.Tensor] = None, static_categorical_features: Optional[torch.Tensor] = None, static_real_features: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> SampleTSPredictionOutput: r""" Greedily generate sequences of sample predictions from a model with a probability distribution head. diff --git a/src/transformers/models/timesfm/modular_timesfm.py b/src/transformers/models/timesfm/modular_timesfm.py index b3fd1c5041eb..b15cae6a7e75 100644 --- a/src/transformers/models/timesfm/modular_timesfm.py +++ b/src/transformers/models/timesfm/modular_timesfm.py @@ -645,8 +645,7 @@ def forward( forecast_context_len: Optional[int] = None, return_forecast_on_context: bool = False, truncate_negative: bool = False, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + ) -> TimesFmOutputForPrediction: r""" window_size (`int`, *optional*): diff --git a/src/transformers/models/timesformer/modeling_timesformer.py b/src/transformers/models/timesformer/modeling_timesformer.py index 19358f0c30e8..7610ab48e3fb 100644 --- a/src/transformers/models/timesformer/modeling_timesformer.py +++ b/src/transformers/models/timesformer/modeling_timesformer.py @@ -509,8 +509,7 @@ class PreTrainedModel def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutput]: r""" @@ -639,8 +638,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: r""" diff --git a/src/transformers/models/timm_backbone/modeling_timm_backbone.py b/src/transformers/models/timm_backbone/modeling_timm_backbone.py index 91e0224866b2..2fab2df53a27 100644 --- a/src/transformers/models/timm_backbone/modeling_timm_backbone.py +++ b/src/transformers/models/timm_backbone/modeling_timm_backbone.py @@ -122,8 +122,7 @@ def _init_weights(self, module): def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[BackboneOutput, tuple[Tensor, ...]]: diff --git a/src/transformers/models/trocr/modeling_trocr.py b/src/transformers/models/trocr/modeling_trocr.py index 11b5b4a415f8..9aa1d28bce03 100644 --- a/src/transformers/models/trocr/modeling_trocr.py +++ b/src/transformers/models/trocr/modeling_trocr.py @@ -768,8 +768,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: r""" diff --git a/src/transformers/models/tvp/modeling_tvp.py b/src/transformers/models/tvp/modeling_tvp.py index 01932573f01b..302a9075c1ab 100644 --- a/src/transformers/models/tvp/modeling_tvp.py +++ b/src/transformers/models/tvp/modeling_tvp.py @@ -495,8 +495,7 @@ def forward( hidden_states, attention_mask=None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): return_dict = return_dict if return_dict is not None else self.config.return_dict @@ -761,8 +760,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ): @@ -864,8 +862,7 @@ def forward( attention_mask: Optional[torch.LongTensor] = None, labels: Optional[tuple[torch.Tensor]] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ): diff --git a/src/transformers/models/udop/modeling_udop.py b/src/transformers/models/udop/modeling_udop.py index 4cca934bcea0..41efd964c99b 100644 --- a/src/transformers/models/udop/modeling_udop.py +++ b/src/transformers/models/udop/modeling_udop.py @@ -1552,8 +1552,7 @@ def forward( decoder_head_mask: Optional[Tensor] = None, cross_attn_head_mask: Optional[Tensor] = None, use_cache=True, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> tuple[Tensor, ...]: @@ -1757,8 +1756,7 @@ def forward( decoder_head_mask: Optional[Tensor] = None, cross_attn_head_mask: Optional[Tensor] = None, use_cache=True, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[Tensor] = None, cache_position: Optional[torch.LongTensor] = None, @@ -1986,8 +1984,7 @@ def forward( visual_bbox: Optional[dict[str, Any]] = None, head_mask: Optional[Tensor] = None, inputs_embeds: Optional[Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutputWithAttentionMask]: r""" diff --git a/src/transformers/models/umt5/modeling_umt5.py b/src/transformers/models/umt5/modeling_umt5.py index d5bb6718baf1..bf0a0a80bbac 100644 --- a/src/transformers/models/umt5/modeling_umt5.py +++ b/src/transformers/models/umt5/modeling_umt5.py @@ -1052,8 +1052,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, decoder_inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqModelOutput]: @@ -1269,8 +1268,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: @@ -1496,8 +1494,7 @@ def forward( attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutput]: r""" @@ -1574,8 +1571,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: r""" @@ -1733,8 +1729,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1852,8 +1847,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], Seq2SeqQuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/unispeech/modular_unispeech.py b/src/transformers/models/unispeech/modular_unispeech.py index e39d5bdd6585..fee5a3b6cc76 100644 --- a/src/transformers/models/unispeech/modular_unispeech.py +++ b/src/transformers/models/unispeech/modular_unispeech.py @@ -249,8 +249,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, UniSpeechBaseModelOutput]: r""" @@ -368,8 +367,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, UniSpeechForPreTrainingOutput]: r""" diff --git a/src/transformers/models/unispeech_sat/modular_unispeech_sat.py b/src/transformers/models/unispeech_sat/modular_unispeech_sat.py index 91565ba4313b..b4c209e58f77 100644 --- a/src/transformers/models/unispeech_sat/modular_unispeech_sat.py +++ b/src/transformers/models/unispeech_sat/modular_unispeech_sat.py @@ -258,8 +258,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, UniSpeechSatBaseModelOutput]: r""" @@ -384,8 +383,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, UniSpeechSatForPreTrainingOutput]: r""" diff --git a/src/transformers/models/upernet/modeling_upernet.py b/src/transformers/models/upernet/modeling_upernet.py index a531fa2abe3f..30dc93a86f42 100644 --- a/src/transformers/models/upernet/modeling_upernet.py +++ b/src/transformers/models/upernet/modeling_upernet.py @@ -305,8 +305,7 @@ def __init__(self, config): def forward( self, pixel_values: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, SemanticSegmenterOutput]: diff --git a/src/transformers/models/video_llava/modeling_video_llava.py b/src/transformers/models/video_llava/modeling_video_llava.py index b945466685f0..ce8f637d0b1c 100644 --- a/src/transformers/models/video_llava/modeling_video_llava.py +++ b/src/transformers/models/video_llava/modeling_video_llava.py @@ -309,8 +309,7 @@ def forward( vision_feature_layer: Optional[Union[int, list[int]]] = None, vision_feature_select_strategy: Optional[str] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -493,8 +492,7 @@ def forward( vision_feature_select_strategy: Optional[str] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/videomae/modeling_videomae.py b/src/transformers/models/videomae/modeling_videomae.py index c418a3b49c54..1c5f49ed7fe3 100755 --- a/src/transformers/models/videomae/modeling_videomae.py +++ b/src/transformers/models/videomae/modeling_videomae.py @@ -541,8 +541,7 @@ def forward( pixel_values: torch.FloatTensor, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -757,8 +756,7 @@ def forward( pixel_values: torch.FloatTensor, bool_masked_pos: torch.BoolTensor, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, VideoMAEForPreTrainingOutput]: r""" @@ -936,8 +934,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: r""" diff --git a/src/transformers/models/vilt/modeling_vilt.py b/src/transformers/models/vilt/modeling_vilt.py index 6ce00d9f397c..974c9c86ae89 100755 --- a/src/transformers/models/vilt/modeling_vilt.py +++ b/src/transformers/models/vilt/modeling_vilt.py @@ -616,8 +616,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, image_token_type_idx: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPooling, tuple[torch.FloatTensor]]: r""" @@ -775,8 +774,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[MaskedLMOutput, tuple[torch.FloatTensor]]: r""" @@ -953,8 +951,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.FloatTensor]]: r""" @@ -1059,8 +1056,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[SequenceClassifierOutput, tuple[torch.FloatTensor]]: r""" @@ -1164,8 +1160,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[ViltForImagesAndTextClassificationOutput, tuple[torch.FloatTensor]]: r""" @@ -1294,8 +1289,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, image_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[TokenClassifierOutput, tuple[torch.FloatTensor]]: r""" diff --git a/src/transformers/models/vipllava/modular_vipllava.py b/src/transformers/models/vipllava/modular_vipllava.py index e9ac97722348..55dd06482b5a 100644 --- a/src/transformers/models/vipllava/modular_vipllava.py +++ b/src/transformers/models/vipllava/modular_vipllava.py @@ -113,8 +113,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, vision_feature_layers: Optional[Union[int, list[int]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **lm_kwargs, @@ -200,8 +199,7 @@ def forward( vision_feature_layers: Optional[Union[int, list[int]]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, diff --git a/src/transformers/models/vision_encoder_decoder/modeling_vision_encoder_decoder.py b/src/transformers/models/vision_encoder_decoder/modeling_vision_encoder_decoder.py index 2202558cfab3..9e4523243dfd 100644 --- a/src/transformers/models/vision_encoder_decoder/modeling_vision_encoder_decoder.py +++ b/src/transformers/models/vision_encoder_decoder/modeling_vision_encoder_decoder.py @@ -446,8 +446,7 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.FloatTensor], Seq2SeqLMOutput]: diff --git a/src/transformers/models/vision_text_dual_encoder/modeling_vision_text_dual_encoder.py b/src/transformers/models/vision_text_dual_encoder/modeling_vision_text_dual_encoder.py index c581807bc443..f870460fbb46 100755 --- a/src/transformers/models/vision_text_dual_encoder/modeling_vision_text_dual_encoder.py +++ b/src/transformers/models/vision_text_dual_encoder/modeling_vision_text_dual_encoder.py @@ -193,8 +193,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, token_type_ids: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], CLIPOutput]: r""" diff --git a/src/transformers/models/visual_bert/modeling_visual_bert.py b/src/transformers/models/visual_bert/modeling_visual_bert.py index ee228e5f4282..8d82926dc651 100755 --- a/src/transformers/models/visual_bert/modeling_visual_bert.py +++ b/src/transformers/models/visual_bert/modeling_visual_bert.py @@ -609,8 +609,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPooling]: r""" @@ -798,8 +797,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, sentence_image_labels: Optional[torch.LongTensor] = None, @@ -943,8 +941,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: @@ -1131,8 +1128,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: @@ -1269,8 +1265,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: @@ -1443,8 +1438,7 @@ def forward( visual_attention_mask: Optional[torch.LongTensor] = None, visual_token_type_ids: Optional[torch.LongTensor] = None, image_text_alignment: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, region_to_phrase_position: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/vit/modeling_vit.py b/src/transformers/models/vit/modeling_vit.py index 6e85320a0fa8..351dac20e3ad 100644 --- a/src/transformers/models/vit/modeling_vit.py +++ b/src/transformers/models/vit/modeling_vit.py @@ -519,8 +519,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -627,8 +626,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedImageModelingOutput]: @@ -747,8 +745,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: diff --git a/src/transformers/models/vit_mae/modeling_vit_mae.py b/src/transformers/models/vit_mae/modeling_vit_mae.py index 61cf29f8569a..f9611dc709ab 100755 --- a/src/transformers/models/vit_mae/modeling_vit_mae.py +++ b/src/transformers/models/vit_mae/modeling_vit_mae.py @@ -685,8 +685,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, noise: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, ViTMAEModelOutput]: @@ -1047,8 +1046,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, noise: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, ViTMAEForPreTrainingOutput]: diff --git a/src/transformers/models/vit_msn/modeling_vit_msn.py b/src/transformers/models/vit_msn/modeling_vit_msn.py index 25efc1ac4de6..0fbfd7e12d99 100644 --- a/src/transformers/models/vit_msn/modeling_vit_msn.py +++ b/src/transformers/models/vit_msn/modeling_vit_msn.py @@ -510,8 +510,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, bool_masked_pos: Optional[torch.BoolTensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: @@ -600,8 +599,7 @@ def forward( pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, ImageClassifierOutput]: diff --git a/src/transformers/models/vitdet/modeling_vitdet.py b/src/transformers/models/vitdet/modeling_vitdet.py index d327650c065a..b216d08c5149 100644 --- a/src/transformers/models/vitdet/modeling_vitdet.py +++ b/src/transformers/models/vitdet/modeling_vitdet.py @@ -681,8 +681,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" diff --git a/src/transformers/models/vitmatte/modeling_vitmatte.py b/src/transformers/models/vitmatte/modeling_vitmatte.py index ff672d106b7c..a264695efbeb 100644 --- a/src/transformers/models/vitmatte/modeling_vitmatte.py +++ b/src/transformers/models/vitmatte/modeling_vitmatte.py @@ -231,8 +231,7 @@ def __init__(self, config): def forward( self, pixel_values: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, ): diff --git a/src/transformers/models/vitpose/modeling_vitpose.py b/src/transformers/models/vitpose/modeling_vitpose.py index e63f69bed634..350c7c848739 100644 --- a/src/transformers/models/vitpose/modeling_vitpose.py +++ b/src/transformers/models/vitpose/modeling_vitpose.py @@ -222,8 +222,7 @@ def forward( dataset_index: Optional[torch.Tensor] = None, flip_pairs: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, VitPoseEstimatorOutput]: r""" diff --git a/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py b/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py index 7579ecb9fbb1..b6880ec4d003 100644 --- a/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py +++ b/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py @@ -461,8 +461,7 @@ def forward( pixel_values: torch.Tensor, dataset_index: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ): r""" diff --git a/src/transformers/models/vits/modeling_vits.py b/src/transformers/models/vits/modeling_vits.py index 65b77a4ccf2f..5f67a0bcb7a7 100644 --- a/src/transformers/models/vits/modeling_vits.py +++ b/src/transformers/models/vits/modeling_vits.py @@ -1119,8 +1119,7 @@ def forward( hidden_states: torch.FloatTensor, padding_mask: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: all_hidden_states = () if output_hidden_states else None @@ -1206,8 +1205,7 @@ def forward( input_ids: torch.Tensor, padding_mask: torch.FloatTensor, attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = True, ) -> Union[tuple[torch.Tensor], VitsTextEncoderOutput]: hidden_states = self.embed_tokens(input_ids) * math.sqrt(self.config.hidden_size) @@ -1307,8 +1305,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, speaker_id: Optional[int] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.FloatTensor] = None, ) -> Union[tuple[Any], VitsModelOutput]: diff --git a/src/transformers/models/vivit/modeling_vivit.py b/src/transformers/models/vivit/modeling_vivit.py index 0617e20de3b9..ab93d5cf6e80 100755 --- a/src/transformers/models/vivit/modeling_vivit.py +++ b/src/transformers/models/vivit/modeling_vivit.py @@ -517,8 +517,7 @@ def forward( self, pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutputWithPooling]: @@ -664,8 +663,7 @@ def forward( pixel_values: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], ImageClassifierOutput]: diff --git a/src/transformers/models/wav2vec2/modeling_wav2vec2.py b/src/transformers/models/wav2vec2/modeling_wav2vec2.py index 9ed86d274e23..b7d764f712cf 100755 --- a/src/transformers/models/wav2vec2/modeling_wav2vec2.py +++ b/src/transformers/models/wav2vec2/modeling_wav2vec2.py @@ -1497,8 +1497,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Wav2Vec2BaseModelOutput]: r""" @@ -1622,8 +1621,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.BoolTensor] = None, sampled_negative_indices: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Wav2Vec2ForPreTrainingOutput]: r""" @@ -1796,8 +1794,7 @@ def forward( self, input_values: torch.FloatTensor, attention_mask: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, MaskedLMOutput]: @@ -1909,8 +1906,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -2032,8 +2028,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: @@ -2147,8 +2142,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -2328,8 +2322,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, XVectorOutput]: diff --git a/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py b/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py index 3427a01808a3..15666f5fa2fc 100644 --- a/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py +++ b/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py @@ -710,8 +710,7 @@ def forward( input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Wav2Vec2BertBaseModelOutput]: r""" @@ -780,8 +779,7 @@ def forward( self, input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -876,8 +874,7 @@ def forward( self, input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: @@ -952,8 +949,7 @@ def forward( input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1017,8 +1013,7 @@ def forward( self, input_features: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, XVectorOutput]: diff --git a/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py b/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py index 70042f7b93b9..dae00494bac3 100644 --- a/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py +++ b/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py @@ -1165,8 +1165,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Wav2Vec2ConformerBaseModelOutput]: r""" @@ -1278,8 +1277,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.BoolTensor] = None, sampled_negative_indices: Optional[torch.BoolTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, Wav2Vec2ConformerForPreTrainingOutput]: r""" @@ -1481,8 +1479,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -1592,8 +1589,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: @@ -1695,8 +1691,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1864,8 +1859,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, XVectorOutput]: diff --git a/src/transformers/models/wavlm/modeling_wavlm.py b/src/transformers/models/wavlm/modeling_wavlm.py index d718d4958a1b..7e2f1d757072 100755 --- a/src/transformers/models/wavlm/modeling_wavlm.py +++ b/src/transformers/models/wavlm/modeling_wavlm.py @@ -1076,8 +1076,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, mask_time_indices: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, WavLMBaseModelOutput]: r""" @@ -1220,8 +1219,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, CausalLMOutput]: @@ -1343,8 +1341,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutput]: @@ -1458,8 +1455,7 @@ def forward( input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1639,8 +1635,7 @@ def forward( self, input_values: Optional[torch.Tensor], attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, labels: Optional[torch.Tensor] = None, ) -> Union[tuple, XVectorOutput]: diff --git a/src/transformers/models/whisper/modeling_whisper.py b/src/transformers/models/whisper/modeling_whisper.py index 613f5fb45a72..3e7f9c082ec3 100644 --- a/src/transformers/models/whisper/modeling_whisper.py +++ b/src/transformers/models/whisper/modeling_whisper.py @@ -1112,8 +1112,7 @@ def forward( decoder_inputs_embeds: Optional[tuple[torch.FloatTensor]] = None, decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: @@ -1282,8 +1281,7 @@ def forward( decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqLMOutput]: @@ -1477,8 +1475,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1617,8 +1614,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, encoder_outputs: Optional[tuple[tuple[torch.FloatTensor]]] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/x_clip/modeling_x_clip.py b/src/transformers/models/x_clip/modeling_x_clip.py index 6c98381a63b8..7b300a9addfa 100644 --- a/src/transformers/models/x_clip/modeling_x_clip.py +++ b/src/transformers/models/x_clip/modeling_x_clip.py @@ -579,8 +579,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -671,8 +670,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -747,8 +745,7 @@ def forward( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -796,8 +793,7 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -892,8 +888,7 @@ def __init__(self, config: XCLIPVisionConfig): def forward( self, pixel_values: torch.FloatTensor, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: @@ -945,8 +940,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -1045,8 +1039,7 @@ def __init__(self, config: XCLIPVisionConfig): def forward( self, hidden_states, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: residual = hidden_states @@ -1221,8 +1214,7 @@ def get_text_features( input_ids: Optional[torch.Tensor] = None, attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1266,8 +1258,7 @@ def get_text_features( def get_video_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1385,8 +1376,7 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, return_loss: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, XCLIPOutput]: diff --git a/src/transformers/models/xglm/modeling_xglm.py b/src/transformers/models/xglm/modeling_xglm.py index 08520e5d3abf..4d1891747f4b 100755 --- a/src/transformers/models/xglm/modeling_xglm.py +++ b/src/transformers/models/xglm/modeling_xglm.py @@ -444,8 +444,7 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: r""" @@ -651,8 +650,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: diff --git a/src/transformers/models/xlm/modeling_xlm.py b/src/transformers/models/xlm/modeling_xlm.py index 73679597944a..8d22f3ddf5ec 100755 --- a/src/transformers/models/xlm/modeling_xlm.py +++ b/src/transformers/models/xlm/modeling_xlm.py @@ -790,8 +790,7 @@ def forward( cache: Optional[dict[str, torch.Tensor]] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # Dummy kwargs for now ) -> Union[tuple, BaseModelOutput]: @@ -1031,8 +1030,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, MaskedLMOutput]: @@ -1124,8 +1122,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1238,8 +1235,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: r""" @@ -1345,8 +1341,7 @@ def forward( is_impossible: Optional[torch.Tensor] = None, cls_index: Optional[torch.Tensor] = None, p_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, XLMForQuestionAnsweringOutput]: r""" @@ -1466,8 +1461,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1557,8 +1551,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" diff --git a/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py b/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py index 323e97a3e534..b435f951a5ab 100644 --- a/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py +++ b/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py @@ -754,8 +754,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -926,8 +925,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1056,8 +1054,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1173,8 +1170,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1268,8 +1264,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1379,8 +1374,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1481,8 +1475,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py b/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py index a0162b8252b6..d24c0f2724b5 100644 --- a/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py +++ b/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py @@ -746,8 +746,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -917,8 +916,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1071,8 +1069,7 @@ def forward( encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -1174,8 +1171,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -1256,8 +1252,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1356,8 +1351,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1453,8 +1447,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/xlnet/modeling_xlnet.py b/src/transformers/models/xlnet/modeling_xlnet.py index ea9f06adeb30..6c6450c25e47 100755 --- a/src/transformers/models/xlnet/modeling_xlnet.py +++ b/src/transformers/models/xlnet/modeling_xlnet.py @@ -1234,8 +1234,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete after depreciation warning is removed ) -> Union[tuple, XLNetModelOutput]: @@ -1562,8 +1561,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetLMHeadModelOutput]: @@ -1747,8 +1745,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetForSequenceClassificationOutput]: @@ -1875,8 +1872,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetForTokenClassificationOutput]: @@ -1985,8 +1981,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetForMultipleChoiceOutput]: @@ -2132,8 +2127,7 @@ def forward( start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetForQuestionAnsweringSimpleOutput]: @@ -2261,8 +2255,7 @@ def forward( cls_index: Optional[torch.Tensor] = None, p_mask: Optional[torch.Tensor] = None, use_mems: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, # delete when `use_cache` is removed in XLNetModel ) -> Union[tuple, XLNetForQuestionAnsweringOutput]: diff --git a/src/transformers/models/xmod/modeling_xmod.py b/src/transformers/models/xmod/modeling_xmod.py index 2d794aa6580f..3d3c3aa4241a 100644 --- a/src/transformers/models/xmod/modeling_xmod.py +++ b/src/transformers/models/xmod/modeling_xmod.py @@ -753,8 +753,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: r""" @@ -909,8 +908,7 @@ def forward( labels: Optional[torch.LongTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], CausalLMOutputWithCrossAttentions]: @@ -1037,8 +1035,7 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MaskedLMOutput]: r""" @@ -1148,8 +1145,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], SequenceClassifierOutput]: r""" @@ -1237,8 +1233,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], MultipleChoiceModelOutput]: r""" @@ -1351,8 +1346,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], TokenClassifierOutput]: r""" @@ -1447,8 +1441,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], QuestionAnsweringModelOutput]: r""" diff --git a/src/transformers/models/yolos/modeling_yolos.py b/src/transformers/models/yolos/modeling_yolos.py index f1a7f4fab822..712232fc2cb3 100755 --- a/src/transformers/models/yolos/modeling_yolos.py +++ b/src/transformers/models/yolos/modeling_yolos.py @@ -587,8 +587,7 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions diff --git a/src/transformers/models/yoso/modeling_yoso.py b/src/transformers/models/yoso/modeling_yoso.py index be705885988e..c00fb97f5189 100644 --- a/src/transformers/models/yoso/modeling_yoso.py +++ b/src/transformers/models/yoso/modeling_yoso.py @@ -696,8 +696,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCrossAttentions]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions @@ -794,8 +793,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MaskedLMOutput]: r""" @@ -885,8 +883,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" @@ -968,8 +965,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, MultipleChoiceModelOutput]: r""" @@ -1075,8 +1071,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, TokenClassifierOutput]: r""" @@ -1153,8 +1148,7 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, start_positions: Optional[torch.Tensor] = None, end_positions: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, QuestionAnsweringModelOutput]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/zamba/modeling_zamba.py b/src/transformers/models/zamba/modeling_zamba.py index ec6803e2d12c..ff69c4defcd2 100644 --- a/src/transformers/models/zamba/modeling_zamba.py +++ b/src/transformers/models/zamba/modeling_zamba.py @@ -925,8 +925,7 @@ def forward( past_key_values: Optional[ZambaHybridDynamicCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: @@ -1103,8 +1102,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, @@ -1276,8 +1274,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: r""" diff --git a/src/transformers/models/zamba2/modular_zamba2.py b/src/transformers/models/zamba2/modular_zamba2.py index 5c5c3a5599b3..99217d1b80a8 100644 --- a/src/transformers/models/zamba2/modular_zamba2.py +++ b/src/transformers/models/zamba2/modular_zamba2.py @@ -1041,8 +1041,7 @@ def forward( past_key_values: Optional[Zamba2HybridDynamicCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: diff --git a/src/transformers/models/zoedepth/modeling_zoedepth.py b/src/transformers/models/zoedepth/modeling_zoedepth.py index eaea8d9dfd68..f1ae92d2ba3f 100644 --- a/src/transformers/models/zoedepth/modeling_zoedepth.py +++ b/src/transformers/models/zoedepth/modeling_zoedepth.py @@ -1269,8 +1269,7 @@ def forward( self, pixel_values: torch.FloatTensor, labels: Optional[torch.LongTensor] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, + return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], DepthEstimatorOutput]: r""" From 4853171a0aa9c4958db2d8524c948dfef1acf55b Mon Sep 17 00:00:00 2001 From: Arthur Date: Wed, 18 Jun 2025 19:43:14 +0900 Subject: [PATCH 33/51] fix merge conflicts with main --- src/transformers/models/biogpt/modular_biogpt.py | 4 ---- src/transformers/models/chameleon/modeling_chameleon.py | 1 - .../models/conditional_detr/modeling_conditional_detr.py | 1 - src/transformers/models/cpmant/modeling_cpmant.py | 1 - src/transformers/models/yolos/modeling_yolos.py | 1 - 5 files changed, 8 deletions(-) diff --git a/src/transformers/models/biogpt/modular_biogpt.py b/src/transformers/models/biogpt/modular_biogpt.py index 22c0549599d5..3431ad03b13d 100644 --- a/src/transformers/models/biogpt/modular_biogpt.py +++ b/src/transformers/models/biogpt/modular_biogpt.py @@ -368,7 +368,6 @@ def forward( past_key_values: Optional[tuple[tuple[torch.Tensor]]] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], @@ -567,7 +566,6 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, **kwargs: Unpack[KwargsForCausalLM], @@ -658,7 +656,6 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, TokenClassifierOutput]: @@ -749,7 +746,6 @@ def forward( labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, position_ids: Optional[torch.LongTensor] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.Tensor] = None, ) -> Union[tuple, SequenceClassifierOutputWithPast]: diff --git a/src/transformers/models/chameleon/modeling_chameleon.py b/src/transformers/models/chameleon/modeling_chameleon.py index 9053e91a30c0..690e2ea4f075 100644 --- a/src/transformers/models/chameleon/modeling_chameleon.py +++ b/src/transformers/models/chameleon/modeling_chameleon.py @@ -944,7 +944,6 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], diff --git a/src/transformers/models/conditional_detr/modeling_conditional_detr.py b/src/transformers/models/conditional_detr/modeling_conditional_detr.py index d6a83cf5a8fb..c39af6b6b337 100644 --- a/src/transformers/models/conditional_detr/modeling_conditional_detr.py +++ b/src/transformers/models/conditional_detr/modeling_conditional_detr.py @@ -1418,7 +1418,6 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], ConditionalDetrModelOutput]: r""" diff --git a/src/transformers/models/cpmant/modeling_cpmant.py b/src/transformers/models/cpmant/modeling_cpmant.py index 5eac5b6dcea3..eb106c962fa1 100755 --- a/src/transformers/models/cpmant/modeling_cpmant.py +++ b/src/transformers/models/cpmant/modeling_cpmant.py @@ -714,7 +714,6 @@ def forward( input_ids: Optional[torch.Tensor] = None, past_key_values: Optional[list[tuple[torch.Tensor, torch.Tensor]]] = None, use_cache: Optional[bool] = None, - labels: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, attention_mask: Optional[torch.Tensor] = None, # dummy parameter for text-generation pipeline diff --git a/src/transformers/models/yolos/modeling_yolos.py b/src/transformers/models/yolos/modeling_yolos.py index 712232fc2cb3..5e7c7daccb64 100755 --- a/src/transformers/models/yolos/modeling_yolos.py +++ b/src/transformers/models/yolos/modeling_yolos.py @@ -587,7 +587,6 @@ def forward( self, pixel_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions From 2ab361bd1d850efd96afb2dbf960a272566e7ac2 Mon Sep 17 00:00:00 2001 From: Arthur Date: Wed, 18 Jun 2025 19:45:24 +0900 Subject: [PATCH 34/51] fix --- src/transformers/models/csm/modular_csm.py | 3 --- src/transformers/models/dab_detr/modeling_dab_detr.py | 3 --- .../models/deformable_detr/modeling_deformable_detr.py | 1 - src/transformers/models/detr/modeling_detr.py | 1 - src/transformers/models/moonshine/modular_moonshine.py | 4 ---- src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py | 1 - src/transformers/models/qwen2_vl/modeling_qwen2_vl.py | 2 -- src/transformers/models/sam/modeling_sam.py | 4 ---- src/transformers/models/superglue/modeling_superglue.py | 1 - .../models/table_transformer/modeling_table_transformer.py | 1 - 10 files changed, 21 deletions(-) diff --git a/src/transformers/models/csm/modular_csm.py b/src/transformers/models/csm/modular_csm.py index 4ec15b5d3dcd..7f91737d01f1 100644 --- a/src/transformers/models/csm/modular_csm.py +++ b/src/transformers/models/csm/modular_csm.py @@ -184,7 +184,6 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: @@ -374,7 +373,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], @@ -683,7 +681,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], diff --git a/src/transformers/models/dab_detr/modeling_dab_detr.py b/src/transformers/models/dab_detr/modeling_dab_detr.py index b1642f88f9fb..eec05bcea0c3 100644 --- a/src/transformers/models/dab_detr/modeling_dab_detr.py +++ b/src/transformers/models/dab_detr/modeling_dab_detr.py @@ -923,7 +923,6 @@ def forward( inputs_embeds, attention_mask, object_queries, - return_dict: Optional[bool] = None, ): r""" @@ -1061,7 +1060,6 @@ def forward( memory_key_padding_mask, object_queries, query_position_embeddings, - return_dict: Optional[bool] = None, ): r""" @@ -1285,7 +1283,6 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DabDetrModelOutput]: r""" diff --git a/src/transformers/models/deformable_detr/modeling_deformable_detr.py b/src/transformers/models/deformable_detr/modeling_deformable_detr.py index 4f7fb40ff2c0..2f361b7b2394 100755 --- a/src/transformers/models/deformable_detr/modeling_deformable_detr.py +++ b/src/transformers/models/deformable_detr/modeling_deformable_detr.py @@ -1544,7 +1544,6 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DeformableDetrModelOutput]: r""" diff --git a/src/transformers/models/detr/modeling_detr.py b/src/transformers/models/detr/modeling_detr.py index e88667a72d4a..d390c9a1a550 100644 --- a/src/transformers/models/detr/modeling_detr.py +++ b/src/transformers/models/detr/modeling_detr.py @@ -1152,7 +1152,6 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], DetrModelOutput]: r""" diff --git a/src/transformers/models/moonshine/modular_moonshine.py b/src/transformers/models/moonshine/modular_moonshine.py index 8faca06f6274..f620e98b1e2f 100644 --- a/src/transformers/models/moonshine/modular_moonshine.py +++ b/src/transformers/models/moonshine/modular_moonshine.py @@ -578,7 +578,6 @@ def forward( self, input_values: Optional[torch.FloatTensor] = None, attention_mask: Optional[torch.Tensor] = None, - **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: r""" @@ -693,7 +692,6 @@ def forward( past_key_values: Optional[Cache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, @@ -834,7 +832,6 @@ def forward( decoder_inputs_embeds: Optional[tuple[torch.FloatTensor]] = None, decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, ) -> Seq2SeqModelOutput: r""" @@ -992,7 +989,6 @@ def forward( decoder_inputs_embeds: Optional[tuple[torch.FloatTensor]] = None, decoder_position_ids: Optional[tuple[torch.LongTensor]] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, labels: Optional[torch.LongTensor] = None, ) -> Seq2SeqLMOutput: diff --git a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py index cf8e391cc03a..adb745929289 100644 --- a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py +++ b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py @@ -613,7 +613,6 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, pixel_values_videos: Optional[torch.FloatTensor] = None, diff --git a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py index a0ac1b0098ce..9bafbdc116f7 100644 --- a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py +++ b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py @@ -884,7 +884,6 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], @@ -1224,7 +1223,6 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, pixel_values: Optional[torch.Tensor] = None, pixel_values_videos: Optional[torch.FloatTensor] = None, diff --git a/src/transformers/models/sam/modeling_sam.py b/src/transformers/models/sam/modeling_sam.py index 733854f43ef1..c9d690d8e03c 100644 --- a/src/transformers/models/sam/modeling_sam.py +++ b/src/transformers/models/sam/modeling_sam.py @@ -1123,7 +1123,6 @@ def get_input_embeddings(self): def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - ) -> SamVisionEncoderOutput: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( @@ -1220,7 +1219,6 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, SamVisionEncoderOutput]: return self.vision_encoder( @@ -1276,7 +1274,6 @@ def get_image_wide_positional_embeddings(self): def get_image_embeddings( self, pixel_values, - ): r""" Returns the image embeddings by passing the pixel values through the vision encoder. @@ -1343,7 +1340,6 @@ def forward( multimask_output: bool = True, attention_similarity: Optional[torch.FloatTensor] = None, target_embedding: Optional[torch.FloatTensor] = None, - **kwargs, ) -> SamImageSegmentationOutput: r""" diff --git a/src/transformers/models/superglue/modeling_superglue.py b/src/transformers/models/superglue/modeling_superglue.py index c34196cbd16c..70b83805b799 100644 --- a/src/transformers/models/superglue/modeling_superglue.py +++ b/src/transformers/models/superglue/modeling_superglue.py @@ -749,7 +749,6 @@ def forward( self, pixel_values: torch.FloatTensor, labels: Optional[torch.LongTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, KeypointMatchingOutput]: r""" diff --git a/src/transformers/models/table_transformer/modeling_table_transformer.py b/src/transformers/models/table_transformer/modeling_table_transformer.py index 51edf1fdefc0..5d5d18ecb22c 100644 --- a/src/transformers/models/table_transformer/modeling_table_transformer.py +++ b/src/transformers/models/table_transformer/modeling_table_transformer.py @@ -1097,7 +1097,6 @@ def forward( encoder_outputs: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], TableTransformerModelOutput]: r""" From 675cd88e1aa842d96cdfc242ec7957b415f51a01 Mon Sep 17 00:00:00 2001 From: Arthur Date: Wed, 18 Jun 2025 22:15:11 +0900 Subject: [PATCH 35/51] proper cleanup of albert --- .../models/albert/modeling_albert.py | 111 +++--------------- 1 file changed, 15 insertions(+), 96 deletions(-) diff --git a/src/transformers/models/albert/modeling_albert.py b/src/transformers/models/albert/modeling_albert.py index fdd3a9bce1af..7ce7f5b2c546 100755 --- a/src/transformers/models/albert/modeling_albert.py +++ b/src/transformers/models/albert/modeling_albert.py @@ -663,15 +663,8 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPooling, tuple]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if input_ids is not None and inputs_embeds is not None: raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") elif input_ids is not None: @@ -700,10 +693,7 @@ def forward( ) use_sdpa_attention_mask = ( - self.attn_implementation == "sdpa" - and self.position_embedding_type == "absolute" - and head_mask is None - and not output_attentions + self.attn_implementation == "sdpa" and self.position_embedding_type == "absolute" and head_mask is None ) if use_sdpa_attention_mask: @@ -721,9 +711,6 @@ def forward( embedding_output, extended_attention_mask, head_mask=head_mask, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) sequence_output = encoder_outputs[0] @@ -736,8 +723,6 @@ def forward( return BaseModelOutputWithPooling( last_hidden_state=sequence_output, pooler_output=pooled_output, - hidden_states=encoder_outputs.hidden_states, - attentions=encoder_outputs.attentions, ) @@ -780,8 +765,7 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, sentence_order_label: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[AlbertForPreTrainingOutput, tuple]: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): @@ -809,7 +793,6 @@ def forward( >>> prediction_logits = outputs.prediction_logits >>> sop_logits = outputs.sop_logits ```""" - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.albert( input_ids, @@ -818,9 +801,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) sequence_output, pooled_output = outputs[:2] @@ -835,16 +816,10 @@ def forward( sentence_order_loss = loss_fct(sop_scores.view(-1, 2), sentence_order_label.view(-1)) total_loss = masked_lm_loss + sentence_order_loss - if not return_dict: - output = (prediction_scores, sop_scores) + outputs[2:] - return ((total_loss,) + output) if total_loss is not None else output - return AlbertForPreTrainingOutput( loss=total_loss, prediction_logits=prediction_scores, sop_logits=sop_scores, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -924,8 +899,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[MaskedLMOutput, tuple]: r""" labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): @@ -962,7 +936,6 @@ def forward( 0.81 ``` """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.albert( input_ids=input_ids, @@ -971,9 +944,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) sequence_outputs = outputs[0] @@ -984,15 +955,9 @@ def forward( loss_fct = CrossEntropyLoss() masked_lm_loss = loss_fct(prediction_scores.view(-1, self.config.vocab_size), labels.view(-1)) - if not return_dict: - output = (prediction_scores,) + outputs[2:] - return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output - return MaskedLMOutput( loss=masked_lm_loss, logits=prediction_scores, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -1025,8 +990,7 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[SequenceClassifierOutput, tuple]: r""" labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): @@ -1034,7 +998,6 @@ def forward( config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.albert( input_ids=input_ids, @@ -1043,9 +1006,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) pooled_output = outputs[1] @@ -1076,15 +1037,9 @@ def forward( loss_fct = BCEWithLogitsLoss() loss = loss_fct(logits, labels) - if not return_dict: - output = (logits,) + outputs[2:] - return ((loss,) + output) if loss is not None else output - return SequenceClassifierOutput( loss=loss, logits=logits, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -1116,15 +1071,8 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[TokenClassifierOutput, tuple]: - r""" - labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): - Labels for computing the token classification loss. Indices should be in `[0, ..., config.num_labels - 1]`. - """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - outputs = self.albert( input_ids, attention_mask=attention_mask, @@ -1132,9 +1080,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) sequence_output = outputs[0] @@ -1147,15 +1093,9 @@ def forward( loss_fct = CrossEntropyLoss() loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1)) - if not return_dict: - output = (logits,) + outputs[2:] - return ((loss,) + output) if loss is not None else output - return TokenClassifierOutput( loss=loss, logits=logits, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -1182,11 +1122,8 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, - ) -> Union[AlbertForPreTrainingOutput, tuple]: - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - + **kwargs, + ) -> Union[QuestionAnsweringModelOutput, tuple]: outputs = self.albert( input_ids=input_ids, attention_mask=attention_mask, @@ -1194,9 +1131,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) sequence_output = outputs[0] @@ -1223,16 +1158,10 @@ def forward( end_loss = loss_fct(end_logits, end_positions) total_loss = (start_loss + end_loss) / 2 - if not return_dict: - output = (start_logits, end_logits) + outputs[2:] - return ((total_loss,) + output) if total_loss is not None else output - return QuestionAnsweringModelOutput( loss=total_loss, start_logits=start_logits, end_logits=end_logits, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -1258,9 +1187,8 @@ def forward( head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, - - return_dict: Optional[bool] = None, - ) -> Union[AlbertForPreTrainingOutput, tuple]: + **kwargs, + ) -> Union[MultipleChoiceModelOutput, tuple]: r""" input_ids (`torch.LongTensor` of shape `(batch_size, num_choices, sequence_length)`): Indices of input sequence tokens in the vocabulary. @@ -1291,7 +1219,6 @@ def forward( num_choices-1]` where *num_choices* is the size of the second dimension of the input tensors. (see *input_ids* above) """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict num_choices = input_ids.shape[1] if input_ids is not None else inputs_embeds.shape[1] input_ids = input_ids.view(-1, input_ids.size(-1)) if input_ids is not None else None @@ -1310,9 +1237,7 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) pooled_output = outputs[1] @@ -1326,15 +1251,9 @@ def forward( loss_fct = CrossEntropyLoss() loss = loss_fct(reshaped_logits, labels) - if not return_dict: - output = (reshaped_logits,) + outputs[2:] - return ((loss,) + output) if loss is not None else output - return MultipleChoiceModelOutput( loss=loss, logits=reshaped_logits, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) From 2dbf9a1757289e08d11f9a8b069849a7544cd6db Mon Sep 17 00:00:00 2001 From: Arthur Date: Wed, 18 Jun 2025 22:38:58 +0900 Subject: [PATCH 36/51] modify align as well --- .../models/align/modeling_align.py | 140 ++---------------- 1 file changed, 15 insertions(+), 125 deletions(-) diff --git a/src/transformers/models/align/modeling_align.py b/src/transformers/models/align/modeling_align.py index f8fe9f658559..416f21fd9ec0 100644 --- a/src/transformers/models/align/modeling_align.py +++ b/src/transformers/models/align/modeling_align.py @@ -504,18 +504,9 @@ def round_repeats(repeats): def forward( self, hidden_states: torch.FloatTensor, - output_hidden_states: Optional[bool] = False, - return_dict: Optional[bool] = True, ) -> BaseModelOutputWithPoolingAndNoAttention: - all_hidden_states = (hidden_states,) if output_hidden_states else None - for block in self.blocks: hidden_states = block(hidden_states) - if output_hidden_states: - all_hidden_states += (hidden_states,) - - if not return_dict: - return tuple(v for v in [hidden_states, all_hidden_states] if v is not None) return BaseModelOutputWithNoAttention( last_hidden_state=hidden_states, @@ -629,7 +620,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: mixed_query_layer = self.query(hidden_states) @@ -715,12 +705,7 @@ def forward( context_layer = context_layer.permute(0, 2, 1, 3).contiguous() new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,) context_layer = context_layer.view(new_context_layer_shape) - - outputs = (context_layer, attention_probs) if output_attentions else (context_layer,) - - if self.is_decoder: - outputs = outputs + (past_key_value,) - return outputs + return context_layer, attention_probs # Copied from transformers.models.bert.modeling_bert.BertSelfOutput with Bert->AlignText @@ -850,7 +835,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: # decoder uni-directional self-attention cached key/values tuple is at positions 1,2 self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None @@ -858,19 +842,14 @@ def forward( hidden_states, attention_mask, head_mask, - output_attentions=output_attentions, past_key_value=self_attn_past_key_value, ) attention_output = self_attention_outputs[0] - - # if decoder, the last output is tuple of self-attn cache if self.is_decoder: outputs = self_attention_outputs[1:-1] - present_key_value = self_attention_outputs[-1] else: outputs = self_attention_outputs[1:] # add self attentions if we output attention weights - cross_attn_present_key_value = None if self.is_decoder and encoder_hidden_states is not None: if not hasattr(self, "crossattention"): raise ValueError( @@ -878,7 +857,6 @@ def forward( " by setting `config.add_cross_attention=True`" ) - # cross_attn cached key/values tuple is at positions 3,4 of past_key_value tuple cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None cross_attention_outputs = self.crossattention( attention_output, @@ -887,25 +865,14 @@ def forward( encoder_hidden_states, encoder_attention_mask, cross_attn_past_key_value, - output_attentions, ) attention_output = cross_attention_outputs[0] outputs = outputs + cross_attention_outputs[1:-1] # add cross attentions if we output attention weights - # add cross-attn cache to positions 3,4 of present_key_value tuple - cross_attn_present_key_value = cross_attention_outputs[-1] - present_key_value = present_key_value + cross_attn_present_key_value - layer_output = apply_chunking_to_forward( self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output ) - outputs = (layer_output,) + outputs - - # if decoder, return the attn key/values as the last output - if self.is_decoder: - outputs = outputs + (present_key_value,) - - return outputs + return layer_output def feed_forward_chunk(self, attention_output): intermediate_output = self.intermediate(attention_output) @@ -929,81 +896,25 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = False, - output_hidden_states: Optional[bool] = False, - return_dict: Optional[bool] = True, - ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: - all_hidden_states = () if output_hidden_states else None - all_self_attentions = () if output_attentions else None - all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None - - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - - next_decoder_cache = () if use_cache else None + **kwargs, + ) -> Union[BaseModelOutputWithPastAndCrossAttentions]: for i, layer_module in enumerate(self.layer): - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - layer_head_mask = head_mask[i] if head_mask is not None else None past_key_value = past_key_values[i] if past_key_values is not None else None - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - layer_module.__call__, - hidden_states, - attention_mask, - layer_head_mask, - encoder_hidden_states, - encoder_attention_mask, - past_key_value, - output_attentions, - ) - else: - layer_outputs = layer_module( - hidden_states, - attention_mask, - layer_head_mask, - encoder_hidden_states, - encoder_attention_mask, - past_key_value, - output_attentions, - ) + layer_outputs = layer_module( + hidden_states, + attention_mask, + layer_head_mask, + encoder_hidden_states, + encoder_attention_mask, + past_key_value, + **kwargs, + ) hidden_states = layer_outputs[0] - if use_cache: - next_decoder_cache += (layer_outputs[-1],) - if output_attentions: - all_self_attentions = all_self_attentions + (layer_outputs[1],) - if self.config.add_cross_attention: - all_cross_attentions = all_cross_attentions + (layer_outputs[2],) - - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - - if not return_dict: - return tuple( - v - for v in [ - hidden_states, - next_decoder_cache, - all_hidden_states, - all_self_attentions, - all_cross_attentions, - ] - if v is not None - ) return BaseModelOutputWithPastAndCrossAttentions( last_hidden_state=hidden_states, - past_key_values=next_decoder_cache, - hidden_states=all_hidden_states, - attentions=all_self_attentions, - cross_attentions=all_cross_attentions, ) @@ -1088,8 +999,7 @@ def forward( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: r""" Examples: @@ -1106,12 +1016,6 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled (EOS token) states ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if input_ids is not None and inputs_embeds is not None: raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") elif input_ids is not None: @@ -1154,25 +1058,13 @@ def forward( inputs_embeds=inputs_embeds, ) encoder_outputs = self.encoder( - embedding_output, - attention_mask=extended_attention_mask, - head_mask=head_mask, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + embedding_output, attention_mask=extended_attention_mask, head_mask=head_mask, **kwargs ) sequence_output = encoder_outputs[0] pooled_output = self.pooler(sequence_output) if self.pooler is not None else None - - if not return_dict: - return (sequence_output, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPoolingAndCrossAttentions( last_hidden_state=sequence_output, pooler_output=pooled_output, - hidden_states=encoder_outputs.hidden_states, - attentions=encoder_outputs.attentions, - cross_attentions=encoder_outputs.cross_attentions, ) @@ -1306,7 +1198,6 @@ def get_text_features( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" @@ -1405,7 +1296,6 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, AlignOutput]: r""" From 6099f095a5148ed7b81ba9cdf95162ef6b1e3608 Mon Sep 17 00:00:00 2001 From: Arthur Date: Wed, 18 Jun 2025 23:28:25 +0900 Subject: [PATCH 37/51] all the files this PR is scoped to tackle --- .../modular-transformers/modeling_dummy.py | 6 +- .../modeling_dummy_bert.py | 6 +- .../modeling_multimodal1.py | 6 +- .../modeling_multimodal2.py | 12 ++-- .../modeling_my_new_model2.py | 6 +- .../modeling_new_task_model.py | 6 +- .../modular-transformers/modeling_roberta.py | 6 +- .../modeling_test_detr.py | 18 ++---- src/transformers/generation/tf_utils.py | 8 +-- .../models/albert/modeling_flax_albert.py | 4 +- .../models/align/modeling_align.py | 20 ++----- .../models/altclip/modeling_altclip.py | 36 ++++------- src/transformers/models/aria/modeling_aria.py | 12 ++-- src/transformers/models/aria/modular_aria.py | 4 +- .../modeling_audio_spectrogram_transformer.py | 6 +- .../models/autoformer/modeling_autoformer.py | 18 ++---- .../models/aya_vision/modeling_aya_vision.py | 12 ++-- .../models/aya_vision/modular_aya_vision.py | 2 +- .../models/bamba/modeling_bamba.py | 6 +- .../models/bamba/modular_bamba.py | 4 +- src/transformers/models/bark/modeling_bark.py | 12 ++-- src/transformers/models/bart/modeling_bart.py | 24 +++----- .../models/bart/modeling_flax_bart.py | 20 ++----- src/transformers/models/beit/modeling_beit.py | 16 ++--- .../models/beit/modeling_flax_beit.py | 4 +- src/transformers/models/bert/modeling_bert.py | 6 +- .../models/bert/modeling_flax_bert.py | 4 +- .../modeling_bert_generation.py | 6 +- .../models/big_bird/modeling_big_bird.py | 6 +- .../models/big_bird/modeling_flax_big_bird.py | 8 +-- .../modeling_bigbird_pegasus.py | 24 +++----- .../models/biogpt/modeling_biogpt.py | 6 +- .../models/biogpt/modular_biogpt.py | 2 +- src/transformers/models/bit/modeling_bit.py | 8 +-- .../models/blenderbot/modeling_blenderbot.py | 24 +++----- .../blenderbot/modeling_flax_blenderbot.py | 16 ++--- .../blenderbot/modeling_tf_blenderbot.py | 4 +- .../modeling_blenderbot_small.py | 24 +++----- .../modeling_flax_blenderbot_small.py | 16 ++--- .../modeling_tf_blenderbot_small.py | 4 +- src/transformers/models/blip/modeling_blip.py | 36 ++++------- .../models/blip/modeling_blip_text.py | 6 +- .../models/blip/modeling_tf_blip.py | 12 +--- .../models/blip/modeling_tf_blip_text.py | 4 +- .../models/blip_2/modeling_blip_2.py | 48 +++++---------- .../models/bloom/modeling_bloom.py | 6 +- .../models/bloom/modeling_flax_bloom.py | 4 +- .../bridgetower/modeling_bridgetower.py | 12 ++-- src/transformers/models/bros/modeling_bros.py | 6 +- .../models/camembert/modeling_camembert.py | 6 +- .../models/canine/modeling_canine.py | 6 +- .../models/chameleon/modeling_chameleon.py | 12 ++-- .../chinese_clip/modeling_chinese_clip.py | 36 ++++------- src/transformers/models/clap/modeling_clap.py | 36 ++++------- src/transformers/models/clip/modeling_clip.py | 42 +++++-------- .../models/clip/modeling_flax_clip.py | 20 ++----- .../models/clipseg/modeling_clipseg.py | 36 ++++------- src/transformers/models/clvp/modeling_clvp.py | 28 +++------ .../models/codegen/modeling_codegen.py | 6 +- .../models/cohere/modeling_cohere.py | 6 +- .../models/cohere/modular_cohere.py | 2 +- .../models/cohere2/modeling_cohere2.py | 12 ++-- .../models/cohere2/modular_cohere2.py | 2 +- .../models/colpali/modeling_colpali.py | 6 +- .../models/colqwen2/modeling_colqwen2.py | 6 +- .../models/colqwen2/modular_colqwen2.py | 2 +- .../modeling_conditional_detr.py | 18 ++---- .../models/convbert/modeling_convbert.py | 6 +- .../models/convnext/modeling_convnext.py | 8 +-- .../models/convnext/modeling_tf_convnext.py | 12 +--- .../models/convnextv2/modeling_convnextv2.py | 8 +-- .../convnextv2/modeling_tf_convnextv2.py | 12 +--- .../models/cpmant/modeling_cpmant.py | 6 +- src/transformers/models/csm/modeling_csm.py | 18 ++---- src/transformers/models/csm/modular_csm.py | 6 +- src/transformers/models/ctrl/modeling_ctrl.py | 6 +- src/transformers/models/cvt/modeling_cvt.py | 4 +- .../models/d_fine/modeling_d_fine.py | 24 +++----- .../models/d_fine/modular_d_fine.py | 2 +- .../models/dab_detr/modeling_dab_detr.py | 24 +++----- .../data2vec/modeling_data2vec_audio.py | 6 +- .../models/data2vec/modeling_data2vec_text.py | 6 +- .../data2vec/modeling_data2vec_vision.py | 10 +--- .../data2vec/modeling_tf_data2vec_vision.py | 8 +-- src/transformers/models/dbrx/modeling_dbrx.py | 12 ++-- .../models/deberta/modeling_deberta.py | 6 +- .../models/deberta_v2/modeling_deberta_v2.py | 6 +- .../modeling_decision_transformer.py | 12 ++-- .../modeling_deformable_detr.py | 18 ++---- src/transformers/models/deit/modeling_deit.py | 6 +- .../models/deit/modeling_tf_deit.py | 4 +- .../models/deprecated/deta/modeling_deta.py | 18 ++---- .../modeling_efficientformer.py | 6 +- .../modeling_tf_efficientformer.py | 4 +- .../deprecated/ernie_m/modeling_ernie_m.py | 6 +- .../models/deprecated/mctct/modeling_mctct.py | 12 ++-- .../models/deprecated/mega/modeling_mega.py | 6 +- .../models/deprecated/mmbt/modeling_mmbt.py | 6 +- .../models/deprecated/nat/modeling_nat.py | 12 ++-- .../models/deprecated/nezha/modeling_nezha.py | 6 +- .../open_llama/modeling_open_llama.py | 12 ++-- .../deprecated/qdqbert/modeling_qdqbert.py | 6 +- .../models/deprecated/realm/modeling_realm.py | 6 +- .../modeling_speech_to_text_2.py | 12 ++-- .../modeling_trajectory_transformer.py | 6 +- .../transfo_xl/modeling_transfo_xl.py | 6 +- .../models/deprecated/tvlt/modeling_tvlt.py | 6 +- .../models/deprecated/van/modeling_van.py | 4 +- .../vit_hybrid/modeling_vit_hybrid.py | 6 +- .../xlm_prophetnet/modeling_xlm_prophetnet.py | 18 ++---- .../depth_anything/modeling_depth_anything.py | 6 +- .../models/depth_pro/modeling_depth_pro.py | 12 ++-- src/transformers/models/detr/modeling_detr.py | 18 ++---- .../models/dinat/modeling_dinat.py | 12 ++-- .../models/dinov2/modeling_dinov2.py | 12 ++-- .../models/dinov2/modeling_flax_dinov2.py | 4 +- .../modeling_dinov2_with_registers.py | 12 ++-- .../modular_dinov2_with_registers.py | 2 +- .../models/distilbert/modeling_distilbert.py | 6 +- .../distilbert/modeling_flax_distilbert.py | 8 +-- .../models/donut/modeling_donut_swin.py | 6 +- src/transformers/models/dpr/modeling_dpr.py | 18 ++---- src/transformers/models/dpt/modeling_dpt.py | 16 ++--- .../efficientnet/modeling_efficientnet.py | 4 +- .../models/electra/modeling_electra.py | 6 +- .../models/electra/modeling_flax_electra.py | 4 +- src/transformers/models/emu3/modeling_emu3.py | 12 ++-- src/transformers/models/emu3/modular_emu3.py | 4 +- .../modeling_flax_encoder_decoder.py | 12 +--- .../models/ernie/modeling_ernie.py | 6 +- src/transformers/models/esm/modeling_esm.py | 6 +- .../models/falcon/modeling_falcon.py | 6 +- .../models/falcon_h1/modeling_falcon_h1.py | 12 ++-- .../models/falcon_h1/modular_falcon_h1.py | 4 +- .../falcon_mamba/modeling_falcon_mamba.py | 4 +- .../modeling_fastspeech2_conformer.py | 10 +--- .../models/flaubert/modeling_flaubert.py | 6 +- .../models/flava/modeling_flava.py | 18 ++---- src/transformers/models/fnet/modeling_fnet.py | 4 +- .../models/focalnet/modeling_focalnet.py | 8 +-- src/transformers/models/fsmt/modeling_fsmt.py | 6 +- .../models/funnel/modeling_funnel.py | 12 ++-- src/transformers/models/fuyu/modeling_fuyu.py | 12 ++-- .../models/gemma/modeling_flax_gemma.py | 4 +- .../models/gemma/modeling_gemma.py | 6 +- .../models/gemma/modular_gemma.py | 2 +- .../models/gemma2/modeling_gemma2.py | 12 ++-- .../models/gemma2/modular_gemma2.py | 4 +- .../models/gemma3/modeling_gemma3.py | 24 +++----- .../models/gemma3/modular_gemma3.py | 6 +- src/transformers/models/git/modeling_git.py | 18 ++---- src/transformers/models/glpn/modeling_glpn.py | 10 +--- .../models/got_ocr2/modeling_got_ocr2.py | 18 ++---- .../models/got_ocr2/modular_got_ocr2.py | 4 +- .../models/gpt2/modeling_flax_gpt2.py | 4 +- src/transformers/models/gpt2/modeling_gpt2.py | 6 +- .../gpt_bigcode/modeling_gpt_bigcode.py | 6 +- .../models/gpt_neo/modeling_flax_gpt_neo.py | 4 +- .../models/gpt_neo/modeling_gpt_neo.py | 6 +- .../models/gpt_neox/modeling_gpt_neox.py | 6 +- .../models/gpt_neox/modular_gpt_neox.py | 2 +- .../modeling_gpt_neox_japanese.py | 6 +- .../models/gptj/modeling_flax_gptj.py | 4 +- src/transformers/models/gptj/modeling_gptj.py | 6 +- .../models/granite/modeling_granite.py | 12 ++-- .../models/granite/modular_granite.py | 4 +- .../granite_speech/modeling_granite_speech.py | 6 +- .../models/granitemoe/modeling_granitemoe.py | 12 ++-- .../modeling_granitemoehybrid.py | 12 ++-- .../modular_granitemoehybrid.py | 2 +- .../modeling_granitemoeshared.py | 12 ++-- .../grounding_dino/modeling_grounding_dino.py | 18 ++---- .../models/groupvit/modeling_groupvit.py | 44 +++++--------- .../models/hgnet_v2/modeling_hgnet_v2.py | 8 +-- .../models/hiera/modeling_hiera.py | 24 +++----- .../models/hubert/modeling_hubert.py | 6 +- .../models/hubert/modular_hubert.py | 2 +- .../models/ibert/modeling_ibert.py | 6 +- .../models/idefics/modeling_idefics.py | 12 ++-- .../models/idefics/modeling_tf_idefics.py | 8 +-- src/transformers/models/idefics/vision.py | 4 +- src/transformers/models/idefics/vision_tf.py | 4 +- .../models/idefics2/modeling_idefics2.py | 24 +++----- .../models/idefics3/modeling_idefics3.py | 24 +++----- .../models/ijepa/modeling_ijepa.py | 6 +- .../models/imagegpt/modeling_imagegpt.py | 6 +- .../models/informer/modeling_informer.py | 18 ++---- .../models/informer/modular_informer.py | 2 +- .../instructblip/modeling_instructblip.py | 18 ++---- .../modeling_instructblipvideo.py | 18 ++---- .../models/internvl/modeling_internvl.py | 18 ++---- .../models/internvl/modular_internvl.py | 4 +- .../models/jamba/modeling_jamba.py | 12 ++-- .../models/janus/modeling_janus.py | 24 +++----- .../models/janus/modular_janus.py | 4 +- .../models/jetmoe/modeling_jetmoe.py | 12 ++-- .../models/kosmos2/modeling_kosmos2.py | 30 ++++------ .../models/layoutlm/modeling_layoutlm.py | 6 +- .../models/layoutlmv2/modeling_layoutlmv2.py | 6 +- .../models/layoutlmv3/modeling_layoutlmv3.py | 6 +- .../layoutlmv3/modeling_tf_layoutlmv3.py | 4 +- src/transformers/models/led/modeling_led.py | 18 ++---- .../models/levit/modeling_levit.py | 4 +- .../models/lightglue/modeling_lightglue.py | 6 +- .../models/lightglue/modular_lightglue.py | 2 +- src/transformers/models/lilt/modeling_lilt.py | 6 +- .../models/llama/modeling_flax_llama.py | 4 +- .../models/llama4/modeling_llama4.py | 30 ++++------ .../models/llava/modeling_llava.py | 12 ++-- .../models/llava_next/modeling_llava_next.py | 12 ++-- .../modeling_llava_next_video.py | 12 ++-- .../modular_llava_next_video.py | 4 +- .../modeling_llava_onevision.py | 12 ++-- .../modular_llava_onevision.py | 4 +- .../models/longformer/modeling_longformer.py | 6 +- .../models/longt5/modeling_flax_longt5.py | 16 ++--- .../models/longt5/modeling_longt5.py | 6 +- src/transformers/models/luke/modeling_luke.py | 6 +- .../models/lxmert/modeling_lxmert.py | 6 +- .../models/m2m_100/modeling_m2m_100.py | 18 ++---- .../models/mamba/modeling_mamba.py | 4 +- .../models/mamba2/modeling_mamba2.py | 4 +- .../models/marian/modeling_flax_marian.py | 16 ++--- .../models/marian/modeling_marian.py | 24 +++----- .../models/marian/modeling_tf_marian.py | 4 +- .../models/markuplm/modeling_markuplm.py | 6 +- .../mask2former/modeling_mask2former.py | 30 ++++------ .../models/maskformer/modeling_maskformer.py | 18 ++---- .../maskformer/modeling_maskformer_swin.py | 12 ++-- .../models/mbart/modeling_flax_mbart.py | 16 ++--- .../models/mbart/modeling_mbart.py | 24 +++----- .../models/mbart/modeling_tf_mbart.py | 4 +- .../megatron_bert/modeling_megatron_bert.py | 6 +- .../models/mgp_str/modeling_mgp_str.py | 12 ++-- src/transformers/models/mimi/modeling_mimi.py | 6 +- .../models/minimax/modeling_minimax.py | 12 ++-- .../models/minimax/modular_minimax.py | 2 +- .../models/mistral/modeling_flax_mistral.py | 4 +- .../models/mistral/modeling_mistral.py | 6 +- .../models/mistral/modular_mistral.py | 2 +- .../models/mistral3/modeling_mistral3.py | 12 ++-- .../models/mistral3/modular_mistral3.py | 4 +- src/transformers/models/mlcd/modeling_mlcd.py | 18 ++---- src/transformers/models/mlcd/modular_mlcd.py | 6 +- .../models/mllama/modeling_mllama.py | 36 ++++------- .../models/mobilebert/modeling_mobilebert.py | 6 +- .../mobilenet_v1/modeling_mobilenet_v1.py | 4 +- .../mobilenet_v2/modeling_mobilenet_v2.py | 8 +-- .../models/mobilevit/modeling_mobilevit.py | 8 +-- .../models/mobilevit/modeling_tf_mobilevit.py | 8 +-- .../mobilevitv2/modeling_mobilevitv2.py | 8 +-- .../models/modernbert/modeling_modernbert.py | 6 +- .../models/modernbert/modular_modernbert.py | 2 +- .../models/moonshine/modeling_moonshine.py | 18 ++---- .../models/moonshine/modular_moonshine.py | 6 +- .../models/moshi/modeling_moshi.py | 18 ++---- .../models/mpnet/modeling_mpnet.py | 6 +- src/transformers/models/mpt/modeling_mpt.py | 6 +- src/transformers/models/mra/modeling_mra.py | 4 +- src/transformers/models/mt5/modeling_mt5.py | 6 +- .../models/musicgen/modeling_musicgen.py | 12 ++-- .../modeling_musicgen_melody.py | 12 ++-- src/transformers/models/mvp/modeling_mvp.py | 24 +++----- .../models/nemotron/modeling_nemotron.py | 12 ++-- .../models/nllb_moe/modeling_nllb_moe.py | 14 ++--- .../nystromformer/modeling_nystromformer.py | 6 +- .../models/olmoe/modeling_olmoe.py | 12 ++-- .../omdet_turbo/modeling_omdet_turbo.py | 22 +++---- .../models/oneformer/modeling_oneformer.py | 24 +++----- .../models/openai/modeling_openai.py | 6 +- .../models/opt/modeling_flax_opt.py | 4 +- src/transformers/models/opt/modeling_opt.py | 18 ++---- .../models/opt/modeling_tf_opt.py | 16 ++--- .../models/owlv2/modeling_owlv2.py | 42 +++++-------- .../models/owlvit/modeling_owlvit.py | 42 +++++-------- .../models/paligemma/modeling_paligemma.py | 12 ++-- .../models/patchtst/modeling_patchtst.py | 12 ++-- .../models/pegasus/modeling_flax_pegasus.py | 16 ++--- .../models/pegasus/modeling_pegasus.py | 24 +++----- .../models/pegasus/modeling_tf_pegasus.py | 4 +- .../models/pegasus_x/modeling_pegasus_x.py | 18 ++---- .../models/perceiver/modeling_perceiver.py | 6 +- .../models/persimmon/modeling_persimmon.py | 12 ++-- src/transformers/models/phi/modeling_phi.py | 6 +- src/transformers/models/phi/modular_phi.py | 2 +- src/transformers/models/phi3/modeling_phi3.py | 6 +- .../modeling_phi4_multimodal.py | 24 +++----- .../modular_phi4_multimodal.py | 6 +- .../models/phimoe/modeling_phimoe.py | 12 ++-- .../models/pix2struct/modeling_pix2struct.py | 12 ++-- .../models/pixtral/modeling_pixtral.py | 6 +- .../models/plbart/modeling_plbart.py | 24 +++----- .../models/plbart/modular_plbart.py | 2 +- .../models/poolformer/modeling_poolformer.py | 4 +- .../models/pop2piano/modeling_pop2piano.py | 6 +- .../modeling_prompt_depth_anything.py | 6 +- .../modular_prompt_depth_anything.py | 2 +- .../models/prophetnet/modeling_prophetnet.py | 18 ++---- src/transformers/models/pvt/modeling_pvt.py | 6 +- .../models/pvt_v2/modeling_pvt_v2.py | 10 +--- .../models/qwen2/modeling_qwen2.py | 6 +- .../models/qwen2/modular_qwen2.py | 2 +- .../qwen2_5_omni/modeling_qwen2_5_omni.py | 24 +++----- .../qwen2_5_omni/modular_qwen2_5_omni.py | 4 +- .../models/qwen2_5_vl/modeling_qwen2_5_vl.py | 18 ++---- .../models/qwen2_5_vl/modular_qwen2_5_vl.py | 4 +- .../qwen2_audio/modeling_qwen2_audio.py | 12 ++-- .../models/qwen2_moe/modeling_qwen2_moe.py | 12 ++-- .../models/qwen2_vl/modeling_qwen2_vl.py | 18 ++---- .../models/qwen3/modeling_qwen3.py | 6 +- .../models/qwen3_moe/modeling_qwen3_moe.py | 6 +- .../models/qwen3_moe/modular_qwen3_moe.py | 2 +- src/transformers/models/rag/modeling_rag.py | 8 +-- .../modeling_recurrent_gemma.py | 8 +-- .../models/reformer/modeling_reformer.py | 6 +- .../models/regnet/modeling_flax_regnet.py | 8 +-- .../models/regnet/modeling_regnet.py | 4 +- .../models/regnet/modeling_tf_regnet.py | 12 +--- .../models/rembert/modeling_rembert.py | 6 +- .../models/resnet/modeling_flax_resnet.py | 8 +-- .../models/resnet/modeling_resnet.py | 8 +-- .../models/resnet/modeling_tf_resnet.py | 8 +-- .../models/roberta/modeling_flax_roberta.py | 4 +- .../models/roberta/modeling_roberta.py | 6 +- .../modeling_flax_roberta_prelayernorm.py | 4 +- .../modeling_roberta_prelayernorm.py | 6 +- .../models/roc_bert/modeling_roc_bert.py | 6 +- .../models/roformer/modeling_flax_roformer.py | 4 +- .../models/roformer/modeling_roformer.py | 6 +- .../models/rt_detr/modeling_rt_detr.py | 24 +++----- .../models/rt_detr/modeling_rt_detr_resnet.py | 4 +- .../models/rt_detr_v2/modeling_rt_detr_v2.py | 24 +++----- src/transformers/models/rwkv/modeling_rwkv.py | 6 +- src/transformers/models/sam/modeling_sam.py | 18 ++---- .../models/sam/modeling_tf_sam.py | 12 +--- .../models/sam_hq/modeling_sam_hq.py | 18 ++---- .../models/sam_hq/modular_sam_hq.py | 4 +- .../seamless_m4t/modeling_seamless_m4t.py | 54 ++++++----------- .../modeling_seamless_m4t_v2.py | 60 +++++++------------ .../models/segformer/modeling_segformer.py | 10 +--- .../models/segformer/modeling_tf_segformer.py | 8 +-- .../models/seggpt/modeling_seggpt.py | 12 ++-- src/transformers/models/sew/modeling_sew.py | 6 +- src/transformers/models/sew/modular_sew.py | 2 +- .../models/sew_d/modeling_sew_d.py | 6 +- .../models/siglip/modeling_siglip.py | 42 +++++-------- .../models/siglip2/modeling_siglip2.py | 42 +++++-------- .../models/siglip2/modular_siglip2.py | 8 +-- .../models/smolvlm/modeling_smolvlm.py | 24 +++----- .../models/smolvlm/modular_smolvlm.py | 2 +- .../modeling_flax_speech_encoder_decoder.py | 12 +--- .../speech_to_text/modeling_speech_to_text.py | 18 ++---- .../modeling_tf_speech_to_text.py | 4 +- .../models/speecht5/modeling_speecht5.py | 18 ++---- .../models/splinter/modeling_splinter.py | 6 +- .../squeezebert/modeling_squeezebert.py | 6 +- .../models/stablelm/modeling_stablelm.py | 12 ++-- .../models/starcoder2/modeling_starcoder2.py | 6 +- .../models/starcoder2/modular_starcoder2.py | 2 +- .../models/superglue/modeling_superglue.py | 6 +- .../models/superpoint/modeling_superpoint.py | 4 +- .../swiftformer/modeling_swiftformer.py | 8 +-- .../swiftformer/modeling_tf_swiftformer.py | 8 +-- src/transformers/models/swin/modeling_swin.py | 12 ++-- .../models/swin/modeling_tf_swin.py | 8 +-- .../models/swin2sr/modeling_swin2sr.py | 6 +- .../models/swinv2/modeling_swinv2.py | 12 ++-- .../modeling_switch_transformers.py | 6 +- .../models/t5/modeling_flax_t5.py | 20 ++----- src/transformers/models/t5/modeling_t5.py | 6 +- .../modeling_table_transformer.py | 18 ++---- .../models/tapas/modeling_tapas.py | 6 +- .../models/textnet/modeling_textnet.py | 8 +-- .../modeling_time_series_transformer.py | 18 ++---- .../timesformer/modeling_timesformer.py | 6 +- .../timm_backbone/modeling_timm_backbone.py | 6 +- .../timm_wrapper/modeling_timm_wrapper.py | 12 ++-- .../models/trocr/modeling_trocr.py | 12 ++-- src/transformers/models/tvp/modeling_tvp.py | 6 +- src/transformers/models/udop/modeling_udop.py | 12 ++-- src/transformers/models/umt5/modeling_umt5.py | 6 +- .../models/unispeech/modeling_unispeech.py | 6 +- .../models/unispeech/modular_unispeech.py | 2 +- .../unispeech_sat/modeling_unispeech_sat.py | 6 +- .../unispeech_sat/modular_unispeech_sat.py | 2 +- .../models/upernet/modeling_upernet.py | 6 +- .../video_llava/modeling_video_llava.py | 12 ++-- .../models/videomae/modeling_videomae.py | 6 +- src/transformers/models/vilt/modeling_vilt.py | 12 ++-- .../models/vipllava/modeling_vipllava.py | 12 ++-- .../models/vipllava/modular_vipllava.py | 4 +- .../modeling_flax_vision_encoder_decoder.py | 12 +--- .../modeling_flax_vision_text_dual_encoder.py | 4 +- .../visual_bert/modeling_visual_bert.py | 6 +- .../models/vit/modeling_flax_vit.py | 4 +- src/transformers/models/vit/modeling_vit.py | 6 +- .../models/vit_mae/modeling_vit_mae.py | 6 +- .../models/vit_msn/modeling_vit_msn.py | 6 +- .../models/vitdet/modeling_vitdet.py | 12 ++-- .../models/vitmatte/modeling_vitmatte.py | 6 +- .../models/vitpose/modeling_vitpose.py | 6 +- .../modeling_vitpose_backbone.py | 6 +- src/transformers/models/vits/modeling_vits.py | 6 +- .../models/vivit/modeling_vivit.py | 6 +- .../models/vjepa2/modeling_vjepa2.py | 6 +- .../models/wav2vec2/modeling_flax_wav2vec2.py | 8 +-- .../models/wav2vec2/modeling_wav2vec2.py | 6 +- .../configuration_wav2vec2_bert.py | 2 +- .../wav2vec2_bert/modeling_wav2vec2_bert.py | 6 +- .../wav2vec2_bert/modular_wav2vec2_bert.py | 2 +- .../modeling_wav2vec2_conformer.py | 6 +- .../models/wavlm/modeling_wavlm.py | 6 +- .../models/whisper/modeling_flax_whisper.py | 24 ++------ .../models/whisper/modeling_tf_whisper.py | 12 +--- .../models/whisper/modeling_whisper.py | 30 ++++------ .../models/x_clip/modeling_x_clip.py | 42 +++++-------- .../models/xglm/modeling_flax_xglm.py | 4 +- .../models/xglm/modeling_tf_xglm.py | 4 +- src/transformers/models/xglm/modeling_xglm.py | 12 ++-- src/transformers/models/xlm/modeling_xlm.py | 6 +- .../xlm_roberta/modeling_flax_xlm_roberta.py | 4 +- .../xlm_roberta/modeling_xlm_roberta.py | 6 +- .../xlm_roberta_xl/modeling_xlm_roberta_xl.py | 6 +- .../models/xlnet/modeling_xlnet.py | 6 +- src/transformers/models/xmod/modeling_xmod.py | 6 +- .../models/yolos/modeling_yolos.py | 6 +- src/transformers/models/yoso/modeling_yoso.py | 6 +- .../models/zamba/modeling_zamba.py | 12 ++-- .../models/zamba2/modeling_zamba2.py | 12 ++-- .../models/zamba2/modular_zamba2.py | 2 +- .../models/zoedepth/modeling_zoedepth.py | 6 +- src/transformers/trainer.py | 4 +- 432 files changed, 1484 insertions(+), 3052 deletions(-) diff --git a/examples/modular-transformers/modeling_dummy.py b/examples/modular-transformers/modeling_dummy.py index ddcf808d9935..cb17ab459b2a 100644 --- a/examples/modular-transformers/modeling_dummy.py +++ b/examples/modular-transformers/modeling_dummy.py @@ -479,10 +479,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/examples/modular-transformers/modeling_dummy_bert.py b/examples/modular-transformers/modeling_dummy_bert.py index 8b2e8aed90bc..fc05912c4260 100644 --- a/examples/modular-transformers/modeling_dummy_bert.py +++ b/examples/modular-transformers/modeling_dummy_bert.py @@ -926,10 +926,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/examples/modular-transformers/modeling_multimodal1.py b/examples/modular-transformers/modeling_multimodal1.py index c16135b6bc42..12dc815f9fb9 100644 --- a/examples/modular-transformers/modeling_multimodal1.py +++ b/examples/modular-transformers/modeling_multimodal1.py @@ -479,10 +479,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/examples/modular-transformers/modeling_multimodal2.py b/examples/modular-transformers/modeling_multimodal2.py index 69e7e454754c..798c917489db 100644 --- a/examples/modular-transformers/modeling_multimodal2.py +++ b/examples/modular-transformers/modeling_multimodal2.py @@ -332,10 +332,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -501,10 +499,8 @@ def forward( Returns: """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values is None: raise ValueError("You have to specify pixel_values") diff --git a/examples/modular-transformers/modeling_my_new_model2.py b/examples/modular-transformers/modeling_my_new_model2.py index d8e10885ef8a..2cd9bad12b68 100644 --- a/examples/modular-transformers/modeling_my_new_model2.py +++ b/examples/modular-transformers/modeling_my_new_model2.py @@ -477,10 +477,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwarg for now ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/examples/modular-transformers/modeling_new_task_model.py b/examples/modular-transformers/modeling_new_task_model.py index 77e4efa172e8..d7c1a48fa463 100644 --- a/examples/modular-transformers/modeling_new_task_model.py +++ b/examples/modular-transformers/modeling_new_task_model.py @@ -291,10 +291,8 @@ def forward( if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict is_training = token_type_ids is not None and labels is not None diff --git a/examples/modular-transformers/modeling_roberta.py b/examples/modular-transformers/modeling_roberta.py index e1bd313a4243..0ac2fdd4e0e3 100644 --- a/examples/modular-transformers/modeling_roberta.py +++ b/examples/modular-transformers/modeling_roberta.py @@ -929,10 +929,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/examples/modular-transformers/modeling_test_detr.py b/examples/modular-transformers/modeling_test_detr.py index de1084de7277..3c7004d0fef6 100644 --- a/examples/modular-transformers/modeling_test_detr.py +++ b/examples/modular-transformers/modeling_test_detr.py @@ -984,10 +984,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1125,10 +1123,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1524,10 +1520,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/generation/tf_utils.py b/src/transformers/generation/tf_utils.py index be51c9cd9f43..2b180922f1ef 100644 --- a/src/transformers/generation/tf_utils.py +++ b/src/transformers/generation/tf_utils.py @@ -1612,7 +1612,7 @@ def greedy_search( eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - output_scores = output_scores if output_scores is not None else self.generation_config.output_scores + output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) @@ -1896,7 +1896,7 @@ def sample( eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - output_scores = output_scores if output_scores is not None else self.generation_config.output_scores + output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) @@ -2241,7 +2241,7 @@ def unflatten_beam_dim(tensor, num_beams, batch_axis=0): output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.generation_config.output_hidden_states ) - output_scores = output_scores if output_scores is not None else self.generation_config.output_scores + return_dict_in_generate = ( return_dict_in_generate if return_dict_in_generate is not None @@ -2772,7 +2772,7 @@ def gather_fn(tensor): eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - output_scores = output_scores if output_scores is not None else self.generation_config.output_scores + output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) diff --git a/src/transformers/models/albert/modeling_flax_albert.py b/src/transformers/models/albert/modeling_flax_albert.py index f2f19cb27716..b4f1d245c2e3 100644 --- a/src/transformers/models/albert/modeling_flax_albert.py +++ b/src/transformers/models/albert/modeling_flax_albert.py @@ -566,9 +566,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/align/modeling_align.py b/src/transformers/models/align/modeling_align.py index 416f21fd9ec0..94c6e72521cb 100644 --- a/src/transformers/models/align/modeling_align.py +++ b/src/transformers/models/align/modeling_align.py @@ -1125,9 +1125,7 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled CLS states ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1217,10 +1215,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1270,9 +1266,7 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1324,10 +1318,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/altclip/modeling_altclip.py b/src/transformers/models/altclip/modeling_altclip.py index ff8d16e8c777..899be3e1d58c 100755 --- a/src/transformers/models/altclip/modeling_altclip.py +++ b/src/transformers/models/altclip/modeling_altclip.py @@ -834,10 +834,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -1034,10 +1032,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1185,10 +1181,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: @@ -1429,10 +1423,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1477,10 +1469,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1532,10 +1522,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( diff --git a/src/transformers/models/aria/modeling_aria.py b/src/transformers/models/aria/modeling_aria.py index 4d81cca0e3dd..a2d677536dcd 100644 --- a/src/transformers/models/aria/modeling_aria.py +++ b/src/transformers/models/aria/modeling_aria.py @@ -1055,10 +1055,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AriaModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is None: @@ -1274,10 +1272,8 @@ def forward( >>> print(generated_texts[1]) Assistant: The bridge is in San Francisco. ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/aria/modular_aria.py b/src/transformers/models/aria/modular_aria.py index 0319e1922b2e..0335deab2f70 100644 --- a/src/transformers/models/aria/modular_aria.py +++ b/src/transformers/models/aria/modular_aria.py @@ -1430,7 +1430,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AriaModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1587,7 +1587,7 @@ def forward( >>> print(generated_texts[1]) Assistant: The bridge is in San Francisco. ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py index 9aa49d1eb2a5..8ab567fa8320 100644 --- a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py +++ b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py @@ -447,10 +447,8 @@ def forward( [`AutoFeatureExtractor`] should be used for extracting the mel features, padding and conversion into a tensor of type `torch.FloatTensor`. See [`~ASTFeatureExtractor.__call__`] """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_values is None: diff --git a/src/transformers/models/autoformer/modeling_autoformer.py b/src/transformers/models/autoformer/modeling_autoformer.py index be15d09293e6..c8d5c5221897 100644 --- a/src/transformers/models/autoformer/modeling_autoformer.py +++ b/src/transformers/models/autoformer/modeling_autoformer.py @@ -1000,10 +1000,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.value_embedding(inputs_embeds) @@ -1177,10 +1175,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1587,10 +1583,8 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/aya_vision/modeling_aya_vision.py b/src/transformers/models/aya_vision/modeling_aya_vision.py index 6430ef425f97..38b056303fb0 100644 --- a/src/transformers/models/aya_vision/modeling_aya_vision.py +++ b/src/transformers/models/aya_vision/modeling_aya_vision.py @@ -285,10 +285,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AyaVisionModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -479,10 +477,8 @@ def forward( >>> gen_tokens = model.generate(**inputs, max_new_tokens=300, do_sample=True, temperature=0.3) >>> processor.tokenizer.decode(gen_tokens[0][inputs.input_ids.shape[1]:], skip_special_tokens=True) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/aya_vision/modular_aya_vision.py b/src/transformers/models/aya_vision/modular_aya_vision.py index 727f917ce551..bc54cc351817 100644 --- a/src/transformers/models/aya_vision/modular_aya_vision.py +++ b/src/transformers/models/aya_vision/modular_aya_vision.py @@ -192,7 +192,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AyaVisionModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/bamba/modeling_bamba.py b/src/transformers/models/bamba/modeling_bamba.py index 3bdeae6b7593..783c8a530208 100644 --- a/src/transformers/models/bamba/modeling_bamba.py +++ b/src/transformers/models/bamba/modeling_bamba.py @@ -1106,10 +1106,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[BambaFlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/bamba/modular_bamba.py b/src/transformers/models/bamba/modular_bamba.py index 303c5feeaf83..83b9655f5b71 100644 --- a/src/transformers/models/bamba/modular_bamba.py +++ b/src/transformers/models/bamba/modular_bamba.py @@ -878,7 +878,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[BambaFlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1135,7 +1135,7 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/bark/modeling_bark.py b/src/transformers/models/bark/modeling_bark.py index 883d88b80fda..6e417eff2ce6 100644 --- a/src/transformers/models/bark/modeling_bark.py +++ b/src/transformers/models/bark/modeling_bark.py @@ -523,10 +523,8 @@ def forward( have to use `input_ids`. If `past_key_values` is not used and `use_cache` is set to `True`, `input_embeds` is used in priority instead of `input_ids`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1189,10 +1187,8 @@ def forward( `past_key_values`). This is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict loss = None diff --git a/src/transformers/models/bart/modeling_bart.py b/src/transformers/models/bart/modeling_bart.py index 7e3156501dee..a0a77668d6df 100755 --- a/src/transformers/models/bart/modeling_bart.py +++ b/src/transformers/models/bart/modeling_bart.py @@ -818,10 +818,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -1030,10 +1028,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1303,10 +1299,8 @@ def forward( input_ids, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1955,10 +1949,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/bart/modeling_flax_bart.py b/src/transformers/models/bart/modeling_flax_bart.py index 818254f3bfa1..07eee3deee08 100644 --- a/src/transformers/models/bart/modeling_flax_bart.py +++ b/src/transformers/models/bart/modeling_flax_bart.py @@ -1026,9 +1026,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1099,9 +1097,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1190,9 +1186,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1366,9 +1360,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1812,9 +1804,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/beit/modeling_beit.py b/src/transformers/models/beit/modeling_beit.py index 78f113222006..37dc69afcb22 100755 --- a/src/transformers/models/beit/modeling_beit.py +++ b/src/transformers/models/beit/modeling_beit.py @@ -817,10 +817,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed @@ -1380,9 +1378,7 @@ def forward( >>> logits = outputs.logits ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and self.config.num_labels == 1: raise ValueError("The number of labels should be greater than one") @@ -1509,10 +1505,8 @@ def forward( [1, 768, 14, 14] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + batch_size = pixel_values.shape[0] embedding_output, (patch_height, patch_width) = self.embeddings(pixel_values) diff --git a/src/transformers/models/beit/modeling_flax_beit.py b/src/transformers/models/beit/modeling_flax_beit.py index c80deace6b39..0cd7e99d1a69 100644 --- a/src/transformers/models/beit/modeling_flax_beit.py +++ b/src/transformers/models/beit/modeling_flax_beit.py @@ -642,9 +642,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/bert/modeling_bert.py b/src/transformers/models/bert/modeling_bert.py index 62c4f41c0f25..bf6181089319 100755 --- a/src/transformers/models/bert/modeling_bert.py +++ b/src/transformers/models/bert/modeling_bert.py @@ -913,10 +913,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/bert/modeling_flax_bert.py b/src/transformers/models/bert/modeling_flax_bert.py index 37828642eb4e..7ea01757a97d 100644 --- a/src/transformers/models/bert/modeling_flax_bert.py +++ b/src/transformers/models/bert/modeling_flax_bert.py @@ -873,9 +873,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/bert_generation/modeling_bert_generation.py b/src/transformers/models/bert_generation/modeling_bert_generation.py index ceab708c0dbc..159a1a5869ff 100755 --- a/src/transformers/models/bert_generation/modeling_bert_generation.py +++ b/src/transformers/models/bert_generation/modeling_bert_generation.py @@ -664,10 +664,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/big_bird/modeling_big_bird.py b/src/transformers/models/big_bird/modeling_big_bird.py index b65d2067adc5..6f380a7a5ddf 100755 --- a/src/transformers/models/big_bird/modeling_big_bird.py +++ b/src/transformers/models/big_bird/modeling_big_bird.py @@ -1906,10 +1906,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs, # NOOP kwargs, for now ) -> Union[BaseModelOutputWithPoolingAndCrossAttentions, tuple[torch.FloatTensor]]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/big_bird/modeling_flax_big_bird.py b/src/transformers/models/big_bird/modeling_flax_big_bird.py index 11dcb30f3d47..4c782e148e16 100644 --- a/src/transformers/models/big_bird/modeling_flax_big_bird.py +++ b/src/transformers/models/big_bird/modeling_flax_big_bird.py @@ -1735,9 +1735,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed @@ -2451,9 +2449,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: diff --git a/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py b/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py index 737c3df2bb5d..1c0f7ab55d8a 100755 --- a/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py +++ b/src/transformers/models/bigbird_pegasus/modeling_bigbird_pegasus.py @@ -1851,10 +1851,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -2194,10 +2192,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2448,10 +2444,8 @@ def forward( input_ids, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3044,10 +3038,8 @@ def forward( >>> logits = outputs.logits ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/biogpt/modeling_biogpt.py b/src/transformers/models/biogpt/modeling_biogpt.py index 440a40f9b6f2..757fc66ed7d8 100755 --- a/src/transformers/models/biogpt/modeling_biogpt.py +++ b/src/transformers/models/biogpt/modeling_biogpt.py @@ -547,10 +547,8 @@ def forward( cache_position: Optional[torch.Tensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/biogpt/modular_biogpt.py b/src/transformers/models/biogpt/modular_biogpt.py index 3431ad03b13d..b79bc180b127 100644 --- a/src/transformers/models/biogpt/modular_biogpt.py +++ b/src/transformers/models/biogpt/modular_biogpt.py @@ -372,7 +372,7 @@ def forward( cache_position: Optional[torch.Tensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/bit/modeling_bit.py b/src/transformers/models/bit/modeling_bit.py index d95e3537819d..6ba233c792ad 100644 --- a/src/transformers/models/bit/modeling_bit.py +++ b/src/transformers/models/bit/modeling_bit.py @@ -674,9 +674,7 @@ def __init__(self, config): def forward( self, pixel_values: Tensor, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None ) -> BaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values) @@ -810,9 +808,7 @@ def forward( >>> outputs = model(**inputs) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + outputs = self.bit(pixel_values, output_hidden_states=True, return_dict=True) diff --git a/src/transformers/models/blenderbot/modeling_blenderbot.py b/src/transformers/models/blenderbot/modeling_blenderbot.py index a8c41689cad8..c1c7fc1f1729 100755 --- a/src/transformers/models/blenderbot/modeling_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_blenderbot.py @@ -771,10 +771,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -987,10 +985,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1256,10 +1252,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 6, 1280] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1611,10 +1605,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/blenderbot/modeling_flax_blenderbot.py b/src/transformers/models/blenderbot/modeling_flax_blenderbot.py index 8b147211881b..8f71a00f2faf 100644 --- a/src/transformers/models/blenderbot/modeling_flax_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_flax_blenderbot.py @@ -999,9 +999,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1074,9 +1072,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1165,9 +1161,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1342,9 +1336,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/blenderbot/modeling_tf_blenderbot.py b/src/transformers/models/blenderbot/modeling_tf_blenderbot.py index 86e576c98280..4cd29be2bfbb 100644 --- a/src/transformers/models/blenderbot/modeling_tf_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_tf_blenderbot.py @@ -1139,9 +1139,7 @@ def call( training=False, **kwargs, ): - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py index f8f57f85bf8a..f8a2ddb85c8d 100755 --- a/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_blenderbot_small.py @@ -757,10 +757,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -968,10 +966,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1223,10 +1219,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 3, 512] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1563,10 +1557,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py index ac30320bbdb4..97044b3439e3 100644 --- a/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py @@ -996,9 +996,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1071,9 +1069,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1161,9 +1157,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1339,9 +1333,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py index 459ef438c53c..aac28c893586 100644 --- a/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py @@ -1140,9 +1140,7 @@ def call( training=False, **kwargs, ): - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/blip/modeling_blip.py b/src/transformers/models/blip/modeling_blip.py index d2c7d143a913..433afb8f1f69 100644 --- a/src/transformers/models/blip/modeling_blip.py +++ b/src/transformers/models/blip/modeling_blip.py @@ -535,10 +535,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -593,10 +591,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -854,10 +850,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use BLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -976,10 +970,8 @@ def forward( ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs = self.vision_model( pixel_values=pixel_values, @@ -1176,10 +1168,8 @@ def forward( ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs = self.vision_model( pixel_values=pixel_values, @@ -1399,10 +1389,8 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs = self.vision_model( pixel_values=pixel_values, diff --git a/src/transformers/models/blip/modeling_blip_text.py b/src/transformers/models/blip/modeling_blip_text.py index 30c7a06beaea..037de4279136 100644 --- a/src/transformers/models/blip/modeling_blip_text.py +++ b/src/transformers/models/blip/modeling_blip_text.py @@ -689,10 +689,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if is_decoder: diff --git a/src/transformers/models/blip/modeling_tf_blip.py b/src/transformers/models/blip/modeling_tf_blip.py index a0bf43892d0e..bec4b5a2c384 100644 --- a/src/transformers/models/blip/modeling_tf_blip.py +++ b/src/transformers/models/blip/modeling_tf_blip.py @@ -650,9 +650,7 @@ def call( Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -733,9 +731,7 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -869,9 +865,7 @@ def call( ) -> Union[tuple, TFBlipOutput]: # Use BLIP model's config for some fields (if specified) instead of those of vision & text components. output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/blip/modeling_tf_blip_text.py b/src/transformers/models/blip/modeling_tf_blip_text.py index e85d3e627ce9..7ed4dd013cae 100644 --- a/src/transformers/models/blip/modeling_tf_blip_text.py +++ b/src/transformers/models/blip/modeling_tf_blip_text.py @@ -853,9 +853,7 @@ def call( `past_key_values`). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if is_decoder: diff --git a/src/transformers/models/blip_2/modeling_blip_2.py b/src/transformers/models/blip_2/modeling_blip_2.py index 3d47b4333e85..9783e72fd85b 100644 --- a/src/transformers/models/blip_2/modeling_blip_2.py +++ b/src/transformers/models/blip_2/modeling_blip_2.py @@ -514,10 +514,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -574,10 +572,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1178,10 +1174,8 @@ def forward( Length of the query, usually based on the number of query tokens. If no value is provided, query_length will be inferred by the query_embeds. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # past_key_values_length @@ -1374,10 +1368,8 @@ def get_text_features( >>> inputs = tokenizer(["a photo of a cat"], padding=True, return_tensors="pt") >>> text_features = model.get_text_features(**inputs) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.use_decoder_only_language_model: @@ -1435,10 +1427,8 @@ def get_image_features( >>> inputs = processor(images=image, return_tensors="pt") >>> image_outputs = model.get_image_features(**inputs) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1480,10 +1470,8 @@ def get_qformer_features( >>> inputs = processor(images=image, return_tensors="pt") >>> qformer_outputs = model.get_qformer_features(**inputs) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1804,10 +1792,8 @@ def forward( >>> print(image_embeds.shape) torch.Size([1, 32, 256]) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -2354,10 +2340,8 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs = self.vision_model( pixel_values=pixel_values, diff --git a/src/transformers/models/bloom/modeling_bloom.py b/src/transformers/models/bloom/modeling_bloom.py index 8b07cc80cbc4..5710a2e765ca 100644 --- a/src/transformers/models/bloom/modeling_bloom.py +++ b/src/transformers/models/bloom/modeling_bloom.py @@ -539,10 +539,8 @@ def forward( if len(deprecated_arguments) > 0: raise ValueError(f"Got unexpected arguments: {deprecated_arguments}") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/bloom/modeling_flax_bloom.py b/src/transformers/models/bloom/modeling_flax_bloom.py index c7bb1cc9c9a5..89f0dab25a52 100644 --- a/src/transformers/models/bloom/modeling_flax_bloom.py +++ b/src/transformers/models/bloom/modeling_flax_bloom.py @@ -472,9 +472,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/bridgetower/modeling_bridgetower.py b/src/transformers/models/bridgetower/modeling_bridgetower.py index 8bc56cd7c922..f871843069fb 100644 --- a/src/transformers/models/bridgetower/modeling_bridgetower.py +++ b/src/transformers/models/bridgetower/modeling_bridgetower.py @@ -1063,10 +1063,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: @@ -1278,10 +1276,8 @@ def forward( >>> outputs.keys() odict_keys(['text_features', 'image_features', 'pooler_output']) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + all_hidden_states_text = () if output_hidden_states else None all_hidden_states_image = () if output_hidden_states else None all_hidden_states_cross = () if output_hidden_states else None diff --git a/src/transformers/models/bros/modeling_bros.py b/src/transformers/models/bros/modeling_bros.py index bcadbdf174cf..1c4e0fd0662a 100755 --- a/src/transformers/models/bros/modeling_bros.py +++ b/src/transformers/models/bros/modeling_bros.py @@ -755,10 +755,8 @@ def forward( >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/camembert/modeling_camembert.py b/src/transformers/models/camembert/modeling_camembert.py index 94344162333c..e4e880dae43f 100644 --- a/src/transformers/models/camembert/modeling_camembert.py +++ b/src/transformers/models/camembert/modeling_camembert.py @@ -826,10 +826,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/canine/modeling_canine.py b/src/transformers/models/canine/modeling_canine.py index 8bd1a5c88477..476ca11ea27d 100644 --- a/src/transformers/models/canine/modeling_canine.py +++ b/src/transformers/models/canine/modeling_canine.py @@ -1020,10 +1020,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, CanineModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + all_hidden_states = () if output_hidden_states else None all_self_attentions = () if output_attentions else None return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/chameleon/modeling_chameleon.py b/src/transformers/models/chameleon/modeling_chameleon.py index 690e2ea4f075..54090d75689a 100644 --- a/src/transformers/models/chameleon/modeling_chameleon.py +++ b/src/transformers/models/chameleon/modeling_chameleon.py @@ -948,10 +948,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1277,10 +1275,8 @@ def forward( >>> generated_ids = model.generate(**inputs, max_new_tokens=100, do_sample=False) >>> processor.batch_decode(generated_ids, skip_special_tokens=True)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs = self.model( diff --git a/src/transformers/models/chinese_clip/modeling_chinese_clip.py b/src/transformers/models/chinese_clip/modeling_chinese_clip.py index 79ba8a059a5b..69a5254571ba 100644 --- a/src/transformers/models/chinese_clip/modeling_chinese_clip.py +++ b/src/transformers/models/chinese_clip/modeling_chinese_clip.py @@ -906,10 +906,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -965,10 +963,8 @@ def forward( interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1066,10 +1062,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: @@ -1283,10 +1277,8 @@ def get_text_features( >>> text_features = text_features / text_features.norm(p=2, dim=-1, keepdim=True) ```""" # Use CHINESE_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1336,10 +1328,8 @@ def get_image_features( >>> image_features = image_features / image_features.norm(p=2, dim=-1, keepdim=True) ```""" # Use CHINESE_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1392,10 +1382,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use CHINESE_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/clap/modeling_clap.py b/src/transformers/models/clap/modeling_clap.py index 9b4e80a81449..f4d25fed3752 100644 --- a/src/transformers/models/clap/modeling_clap.py +++ b/src/transformers/models/clap/modeling_clap.py @@ -1627,10 +1627,8 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return self.audio_encoder( input_features=input_features, @@ -1696,10 +1694,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: @@ -1853,10 +1849,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use CLAP model's config for some fields (if specified) instead of those of audio & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1907,10 +1901,8 @@ def get_audio_features( >>> inputs = feature_extractor(random_audio, return_tensors="pt") >>> audio_features = model.get_audio_features(**inputs) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict audio_outputs = self.audio_model( @@ -1969,10 +1961,8 @@ def forward( >>> probs = logits_per_audio.softmax(dim=-1) # we can take the softmax to get the label probabilities ```""" # Use CLAP model's config for some fields (if specified) instead of those of audio & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict audio_outputs = self.audio_model( @@ -2144,10 +2134,8 @@ def forward( >>> audio_embeds = outputs.audio_embeds ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + audio_outputs = self.audio_model( input_features=input_features, diff --git a/src/transformers/models/clip/modeling_clip.py b/src/transformers/models/clip/modeling_clip.py index bf6bf85b5b30..f0c29823a585 100644 --- a/src/transformers/models/clip/modeling_clip.py +++ b/src/transformers/models/clip/modeling_clip.py @@ -562,10 +562,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -629,10 +627,8 @@ def forward( position_ids: Optional[torch.Tensor] = None, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if input_ids is None: raise ValueError("You have to specify input_ids") @@ -769,10 +765,8 @@ def forward( interpolate_pos_encoding: Optional[bool] = False, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -919,10 +913,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + text_outputs: BaseModelOutputWithPooling = self.text_model( input_ids=input_ids, @@ -967,10 +959,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1022,10 +1012,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1240,10 +1228,8 @@ def forward( config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values, diff --git a/src/transformers/models/clip/modeling_flax_clip.py b/src/transformers/models/clip/modeling_flax_clip.py index 0394974d0647..98102c0369c9 100644 --- a/src/transformers/models/clip/modeling_flax_clip.py +++ b/src/transformers/models/clip/modeling_flax_clip.py @@ -532,9 +532,7 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(input_ids=input_ids, position_ids=position_ids) @@ -594,9 +592,7 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(pixel_values) @@ -675,9 +671,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: @@ -753,9 +747,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -831,9 +823,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: diff --git a/src/transformers/models/clipseg/modeling_clipseg.py b/src/transformers/models/clipseg/modeling_clipseg.py index 6b81fdafe563..98870c860fdd 100644 --- a/src/transformers/models/clipseg/modeling_clipseg.py +++ b/src/transformers/models/clipseg/modeling_clipseg.py @@ -525,10 +525,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -590,10 +588,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None: @@ -730,10 +726,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = True, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(pixel_values, interpolate_pos_encoding=interpolate_pos_encoding) @@ -873,10 +867,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use CLIPSEG model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -924,10 +916,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use CLIPSEG model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -981,10 +971,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use CLIPSEG model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/clvp/modeling_clvp.py b/src/transformers/models/clvp/modeling_clvp.py index 3d072cea61a6..3d69ebe817c5 100644 --- a/src/transformers/models/clvp/modeling_clvp.py +++ b/src/transformers/models/clvp/modeling_clvp.py @@ -913,10 +913,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: @@ -1047,10 +1045,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1209,10 +1205,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1409,10 +1403,8 @@ def forward( are ignored (masked), the loss is only computed for labels in `[0, ..., config.vocab_size]` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1751,9 +1743,7 @@ def forward( """ # Use CLVP model's config for some fields (if specified) instead of those of speech & text components. - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict conditioning_embeds = self.conditioning_encoder( diff --git a/src/transformers/models/codegen/modeling_codegen.py b/src/transformers/models/codegen/modeling_codegen.py index 9401f419ac12..d6c1cd747fe6 100644 --- a/src/transformers/models/codegen/modeling_codegen.py +++ b/src/transformers/models/codegen/modeling_codegen.py @@ -368,10 +368,8 @@ def forward( is useful if you want more control over how to convert *input_ids* indices into associated vectors than the model's internal embedding lookup matrix. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/cohere/modeling_cohere.py b/src/transformers/models/cohere/modeling_cohere.py index 744c5a217161..5f8f6b353265 100644 --- a/src/transformers/models/cohere/modeling_cohere.py +++ b/src/transformers/models/cohere/modeling_cohere.py @@ -538,10 +538,8 @@ def forward( >> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/cohere/modular_cohere.py b/src/transformers/models/cohere/modular_cohere.py index c5e5b957ec99..d6d55192a012 100644 --- a/src/transformers/models/cohere/modular_cohere.py +++ b/src/transformers/models/cohere/modular_cohere.py @@ -342,7 +342,7 @@ def forward( >> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/cohere2/modeling_cohere2.py b/src/transformers/models/cohere2/modeling_cohere2.py index 48b90383f824..b7b02a7357aa 100644 --- a/src/transformers/models/cohere2/modeling_cohere2.py +++ b/src/transformers/models/cohere2/modeling_cohere2.py @@ -396,10 +396,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -561,10 +559,8 @@ def forward( >> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/cohere2/modular_cohere2.py b/src/transformers/models/cohere2/modular_cohere2.py index 5838b09e19b0..91b8bb50f1fe 100644 --- a/src/transformers/models/cohere2/modular_cohere2.py +++ b/src/transformers/models/cohere2/modular_cohere2.py @@ -416,7 +416,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/colpali/modeling_colpali.py b/src/transformers/models/colpali/modeling_colpali.py index b108bea2adfd..f6e413a21f8f 100644 --- a/src/transformers/models/colpali/modeling_colpali.py +++ b/src/transformers/models/colpali/modeling_colpali.py @@ -136,11 +136,9 @@ def forward( ) -> ColPaliForRetrievalOutput: if pixel_values is not None: pixel_values = pixel_values.to(dtype=self.dtype) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vlm_output = self.vlm( diff --git a/src/transformers/models/colqwen2/modeling_colqwen2.py b/src/transformers/models/colqwen2/modeling_colqwen2.py index a38746fe12dd..458ca75cedd7 100644 --- a/src/transformers/models/colqwen2/modeling_colqwen2.py +++ b/src/transformers/models/colqwen2/modeling_colqwen2.py @@ -164,11 +164,9 @@ def forward( dim=0, ) # (num_patches_h * num_patches_w, pixel_values) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict position_ids, rope_deltas = self.vlm.model.get_rope_index( diff --git a/src/transformers/models/colqwen2/modular_colqwen2.py b/src/transformers/models/colqwen2/modular_colqwen2.py index 892a2a687009..b0f8efc61a4d 100644 --- a/src/transformers/models/colqwen2/modular_colqwen2.py +++ b/src/transformers/models/colqwen2/modular_colqwen2.py @@ -315,7 +315,7 @@ def forward( dim=0, ) # (num_patches_h * num_patches_w, pixel_values) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states diff --git a/src/transformers/models/conditional_detr/modeling_conditional_detr.py b/src/transformers/models/conditional_detr/modeling_conditional_detr.py index c39af6b6b337..efbdb119b8c3 100644 --- a/src/transformers/models/conditional_detr/modeling_conditional_detr.py +++ b/src/transformers/models/conditional_detr/modeling_conditional_detr.py @@ -1116,10 +1116,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1250,10 +1248,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1455,10 +1451,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/convbert/modeling_convbert.py b/src/transformers/models/convbert/modeling_convbert.py index 7ef942ea3058..7716b81a7ad1 100755 --- a/src/transformers/models/convbert/modeling_convbert.py +++ b/src/transformers/models/convbert/modeling_convbert.py @@ -819,10 +819,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/convnext/modeling_convnext.py b/src/transformers/models/convnext/modeling_convnext.py index ccb784510310..59bb0683f9c8 100755 --- a/src/transformers/models/convnext/modeling_convnext.py +++ b/src/transformers/models/convnext/modeling_convnext.py @@ -296,9 +296,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -452,9 +450,7 @@ def forward( >>> outputs = model(**inputs) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/convnext/modeling_tf_convnext.py b/src/transformers/models/convnext/modeling_tf_convnext.py index d2e5b0c6f4a6..a91be6221e47 100644 --- a/src/transformers/models/convnext/modeling_tf_convnext.py +++ b/src/transformers/models/convnext/modeling_tf_convnext.py @@ -369,9 +369,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor]]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -527,9 +525,7 @@ def call( >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -623,9 +619,7 @@ def call( >>> predicted_class_idx = tf.math.argmax(logits, axis=-1)[0] >>> print("Predicted class:", model.config.id2label[int(predicted_class_idx)]) ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/convnextv2/modeling_convnextv2.py b/src/transformers/models/convnextv2/modeling_convnextv2.py index ef9d282b9ef0..a3b0e1b552ab 100644 --- a/src/transformers/models/convnextv2/modeling_convnextv2.py +++ b/src/transformers/models/convnextv2/modeling_convnextv2.py @@ -317,9 +317,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -475,9 +473,7 @@ def forward( >>> outputs = model(**inputs) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/convnextv2/modeling_tf_convnextv2.py b/src/transformers/models/convnextv2/modeling_tf_convnextv2.py index 9c2484a1ba75..70f594c95ae2 100644 --- a/src/transformers/models/convnextv2/modeling_tf_convnextv2.py +++ b/src/transformers/models/convnextv2/modeling_tf_convnextv2.py @@ -415,9 +415,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor]]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -561,9 +559,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPoolingAndNoAttention, tuple[tf.Tensor]]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -638,9 +634,7 @@ def call( config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/cpmant/modeling_cpmant.py b/src/transformers/models/cpmant/modeling_cpmant.py index eb106c962fa1..da9be8d45887 100755 --- a/src/transformers/models/cpmant/modeling_cpmant.py +++ b/src/transformers/models/cpmant/modeling_cpmant.py @@ -603,10 +603,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict use_cache = use_cache if use_cache is not None else self.config.use_cache diff --git a/src/transformers/models/csm/modeling_csm.py b/src/transformers/models/csm/modeling_csm.py index a2c713359bf8..00f4367821c5 100644 --- a/src/transformers/models/csm/modeling_csm.py +++ b/src/transformers/models/csm/modeling_csm.py @@ -458,10 +458,8 @@ def forward( ) position_ids = None - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -637,10 +635,8 @@ def forward( config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs = self.model( @@ -1100,10 +1096,8 @@ def forward( >>> output.loss.backward() ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if input_ids is not None and input_ids.ndim == 2: merged_inputs = self._merge_input_ids_with_input_values( diff --git a/src/transformers/models/csm/modular_csm.py b/src/transformers/models/csm/modular_csm.py index 7f91737d01f1..794e0ea32c4b 100644 --- a/src/transformers/models/csm/modular_csm.py +++ b/src/transformers/models/csm/modular_csm.py @@ -199,7 +199,7 @@ def forward( ) position_ids = None - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -386,7 +386,7 @@ def forward( config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -752,7 +752,7 @@ def forward( >>> output.loss.backward() ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/ctrl/modeling_ctrl.py b/src/transformers/models/ctrl/modeling_ctrl.py index b6c33789ad2f..5856631f0df5 100644 --- a/src/transformers/models/ctrl/modeling_ctrl.py +++ b/src/transformers/models/ctrl/modeling_ctrl.py @@ -309,11 +309,9 @@ def forward( >>> list(last_hidden_states.shape) [1, 5, 1280] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + use_cache = use_cache if use_cache is not None else self.config.use_cache - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/cvt/modeling_cvt.py b/src/transformers/models/cvt/modeling_cvt.py index d0cd4b554b17..f7cf2ab62072 100644 --- a/src/transformers/models/cvt/modeling_cvt.py +++ b/src/transformers/models/cvt/modeling_cvt.py @@ -561,9 +561,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCLSToken]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/d_fine/modeling_d_fine.py b/src/transformers/models/d_fine/modeling_d_fine.py index b2c276190921..e8675560349b 100644 --- a/src/transformers/models/d_fine/modeling_d_fine.py +++ b/src/transformers/models/d_fine/modeling_d_fine.py @@ -1184,10 +1184,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1482,10 +1480,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape @@ -1774,10 +1770,8 @@ def forward( Detected remote with confidence 0.947 at location [40.46, 73.7, 175.62, 117.57] Detected sofa with confidence 0.918 at location [0.59, 1.88, 640.25, 474.74] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( @@ -2164,10 +2158,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds diff --git a/src/transformers/models/d_fine/modular_d_fine.py b/src/transformers/models/d_fine/modular_d_fine.py index 03791838c27e..24d10023ea60 100644 --- a/src/transformers/models/d_fine/modular_d_fine.py +++ b/src/transformers/models/d_fine/modular_d_fine.py @@ -739,7 +739,7 @@ def forward( output_attentions=None, return_dict=None, ) -> DFineDecoderOutput: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/dab_detr/modeling_dab_detr.py b/src/transformers/models/dab_detr/modeling_dab_detr.py index eec05bcea0c3..fcc0bcfb230c 100644 --- a/src/transformers/models/dab_detr/modeling_dab_detr.py +++ b/src/transformers/models/dab_detr/modeling_dab_detr.py @@ -950,10 +950,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1085,10 +1083,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1320,10 +1316,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, _, height, width = pixel_values.shape @@ -1588,10 +1582,8 @@ def forward( Detected remote with confidence 0.683 at location [334.48, 73.49, 366.37, 190.01] Detected couch with confidence 0.535 at location [0.52, 1.19, 640.35, 475.1] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # First, sent images through DAB_DETR base model to obtain encoder + decoder outputs diff --git a/src/transformers/models/data2vec/modeling_data2vec_audio.py b/src/transformers/models/data2vec/modeling_data2vec_audio.py index 7605c9f980c2..59f7f51502cc 100755 --- a/src/transformers/models/data2vec/modeling_data2vec_audio.py +++ b/src/transformers/models/data2vec/modeling_data2vec_audio.py @@ -830,10 +830,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/data2vec/modeling_data2vec_text.py b/src/transformers/models/data2vec/modeling_data2vec_text.py index 612d2d80fb17..61dcd9360bc4 100644 --- a/src/transformers/models/data2vec/modeling_data2vec_text.py +++ b/src/transformers/models/data2vec/modeling_data2vec_text.py @@ -661,10 +661,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/data2vec/modeling_data2vec_vision.py b/src/transformers/models/data2vec/modeling_data2vec_vision.py index c5881c7b39aa..e9a4646b23ab 100644 --- a/src/transformers/models/data2vec/modeling_data2vec_vision.py +++ b/src/transformers/models/data2vec/modeling_data2vec_vision.py @@ -823,10 +823,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed @@ -1302,9 +1300,7 @@ def forward( >>> logits = outputs.logits ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and self.config.num_labels == 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py b/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py index 399da1d9d838..fcd7e81c3abd 100644 --- a/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py +++ b/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py @@ -789,9 +789,7 @@ def call( training: bool = False, ) -> Union[tuple, TFData2VecVisionModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1629,9 +1627,7 @@ def call( >>> logits = outputs.logits ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and self.config.num_labels == 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/dbrx/modeling_dbrx.py b/src/transformers/models/dbrx/modeling_dbrx.py index 29d9b4501ad6..2aaf6ce0c600 100644 --- a/src/transformers/models/dbrx/modeling_dbrx.py +++ b/src/transformers/models/dbrx/modeling_dbrx.py @@ -877,10 +877,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, MoeModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) @@ -1202,10 +1200,8 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) diff --git a/src/transformers/models/deberta/modeling_deberta.py b/src/transformers/models/deberta/modeling_deberta.py index a77aa106a8f0..7e0272dd4275 100644 --- a/src/transformers/models/deberta/modeling_deberta.py +++ b/src/transformers/models/deberta/modeling_deberta.py @@ -683,10 +683,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/deberta_v2/modeling_deberta_v2.py b/src/transformers/models/deberta_v2/modeling_deberta_v2.py index 38674e1cf563..0d17d47aa039 100644 --- a/src/transformers/models/deberta_v2/modeling_deberta_v2.py +++ b/src/transformers/models/deberta_v2/modeling_deberta_v2.py @@ -761,10 +761,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/decision_transformer/modeling_decision_transformer.py b/src/transformers/models/decision_transformer/modeling_decision_transformer.py index f24ad353cd30..3fe4cdd9db6b 100755 --- a/src/transformers/models/decision_transformer/modeling_decision_transformer.py +++ b/src/transformers/models/decision_transformer/modeling_decision_transformer.py @@ -530,10 +530,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -885,10 +883,8 @@ def forward( ... ) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, seq_length = states.shape[0], states.shape[1] diff --git a/src/transformers/models/deformable_detr/modeling_deformable_detr.py b/src/transformers/models/deformable_detr/modeling_deformable_detr.py index 2f361b7b2394..88c8df05a59b 100755 --- a/src/transformers/models/deformable_detr/modeling_deformable_detr.py +++ b/src/transformers/models/deformable_detr/modeling_deformable_detr.py @@ -1109,10 +1109,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1243,10 +1241,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1577,10 +1573,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/deit/modeling_deit.py b/src/transformers/models/deit/modeling_deit.py index a732b8609dbc..cc2b463665d2 100644 --- a/src/transformers/models/deit/modeling_deit.py +++ b/src/transformers/models/deit/modeling_deit.py @@ -520,10 +520,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deit/modeling_tf_deit.py b/src/transformers/models/deit/modeling_tf_deit.py index 6dc29ae7fb04..b45400f8b34e 100644 --- a/src/transformers/models/deit/modeling_tf_deit.py +++ b/src/transformers/models/deit/modeling_tf_deit.py @@ -637,9 +637,7 @@ def call( training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deprecated/deta/modeling_deta.py b/src/transformers/models/deprecated/deta/modeling_deta.py index ae44b5e1b12f..bb114858ccb1 100644 --- a/src/transformers/models/deprecated/deta/modeling_deta.py +++ b/src/transformers/models/deprecated/deta/modeling_deta.py @@ -1196,10 +1196,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1312,10 +1310,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1627,10 +1623,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 900, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/deprecated/efficientformer/modeling_efficientformer.py b/src/transformers/models/deprecated/efficientformer/modeling_efficientformer.py index fa84afc00d3c..1f8ee1dcc377 100644 --- a/src/transformers/models/deprecated/efficientformer/modeling_efficientformer.py +++ b/src/transformers/models/deprecated/efficientformer/modeling_efficientformer.py @@ -574,10 +574,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py b/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py index 12d7c942ae50..7885287d27b5 100644 --- a/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py +++ b/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py @@ -835,9 +835,7 @@ def call( ) -> Union[TFBaseModelOutput, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deprecated/ernie_m/modeling_ernie_m.py b/src/transformers/models/deprecated/ernie_m/modeling_ernie_m.py index 9f0d8e676ada..e7fd1c294c75 100755 --- a/src/transformers/models/deprecated/ernie_m/modeling_ernie_m.py +++ b/src/transformers/models/deprecated/ernie_m/modeling_ernie_m.py @@ -529,10 +529,8 @@ def forward( raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time.") # init the default bool value - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers) diff --git a/src/transformers/models/deprecated/mctct/modeling_mctct.py b/src/transformers/models/deprecated/mctct/modeling_mctct.py index e4852ff78f87..a0de093c0478 100755 --- a/src/transformers/models/deprecated/mctct/modeling_mctct.py +++ b/src/transformers/models/deprecated/mctct/modeling_mctct.py @@ -548,10 +548,8 @@ def forward( output_hidden_states: bool = False, return_dict: bool = True, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict input_features = self.layer_norm(input_features) @@ -655,10 +653,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_features is None: diff --git a/src/transformers/models/deprecated/mega/modeling_mega.py b/src/transformers/models/deprecated/mega/modeling_mega.py index d6ac15bebf79..756e2c54c53d 100644 --- a/src/transformers/models/deprecated/mega/modeling_mega.py +++ b/src/transformers/models/deprecated/mega/modeling_mega.py @@ -1520,10 +1520,8 @@ def forward( `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/deprecated/mmbt/modeling_mmbt.py b/src/transformers/models/deprecated/mmbt/modeling_mmbt.py index 45ae577f7fce..e45c49171426 100644 --- a/src/transformers/models/deprecated/mmbt/modeling_mmbt.py +++ b/src/transformers/models/deprecated/mmbt/modeling_mmbt.py @@ -216,10 +216,8 @@ def forward( encoder = ImageEncoder(args) mmbt = MMBTModel(config, transformer, encoder) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/deprecated/nat/modeling_nat.py b/src/transformers/models/deprecated/nat/modeling_nat.py index 930c8f298211..db0f122ca8e8 100644 --- a/src/transformers/models/deprecated/nat/modeling_nat.py +++ b/src/transformers/models/deprecated/nat/modeling_nat.py @@ -710,10 +710,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, NatModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -908,10 +906,8 @@ def forward( [1, 512, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/deprecated/nezha/modeling_nezha.py b/src/transformers/models/deprecated/nezha/modeling_nezha.py index d1c3fd8dbaa5..78b9c3bec997 100644 --- a/src/transformers/models/deprecated/nezha/modeling_nezha.py +++ b/src/transformers/models/deprecated/nezha/modeling_nezha.py @@ -911,10 +911,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/deprecated/open_llama/modeling_open_llama.py b/src/transformers/models/deprecated/open_llama/modeling_open_llama.py index 100d02e22855..4141536688df 100644 --- a/src/transformers/models/deprecated/open_llama/modeling_open_llama.py +++ b/src/transformers/models/deprecated/open_llama/modeling_open_llama.py @@ -562,10 +562,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -746,10 +744,8 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/deprecated/qdqbert/modeling_qdqbert.py b/src/transformers/models/deprecated/qdqbert/modeling_qdqbert.py index f30a07570095..a6588bb2dc2f 100755 --- a/src/transformers/models/deprecated/qdqbert/modeling_qdqbert.py +++ b/src/transformers/models/deprecated/qdqbert/modeling_qdqbert.py @@ -898,10 +898,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/deprecated/realm/modeling_realm.py b/src/transformers/models/deprecated/realm/modeling_realm.py index 5714bf52a0ec..acf7c592898a 100644 --- a/src/transformers/models/deprecated/realm/modeling_realm.py +++ b/src/transformers/models/deprecated/realm/modeling_realm.py @@ -1033,10 +1033,8 @@ def forward( output_hidden_states=None, return_dict=None, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/deprecated/speech_to_text_2/modeling_speech_to_text_2.py b/src/transformers/models/deprecated/speech_to_text_2/modeling_speech_to_text_2.py index ce4fdd1bb2a0..12e59d8d3f4f 100755 --- a/src/transformers/models/deprecated/speech_to_text_2/modeling_speech_to_text_2.py +++ b/src/transformers/models/deprecated/speech_to_text_2/modeling_speech_to_text_2.py @@ -539,10 +539,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -849,10 +847,8 @@ def forward( >>> loss.backward() # doctest: +IGNORE_RESULT ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/deprecated/trajectory_transformer/modeling_trajectory_transformer.py b/src/transformers/models/deprecated/trajectory_transformer/modeling_trajectory_transformer.py index a06a52d9f3cc..86445d8043b3 100644 --- a/src/transformers/models/deprecated/trajectory_transformer/modeling_trajectory_transformer.py +++ b/src/transformers/models/deprecated/trajectory_transformer/modeling_trajectory_transformer.py @@ -504,10 +504,8 @@ def forward( ... ) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if past_key_values is None: past_key_values = tuple([None] * len(self.blocks)) diff --git a/src/transformers/models/deprecated/transfo_xl/modeling_transfo_xl.py b/src/transformers/models/deprecated/transfo_xl/modeling_transfo_xl.py index 664bc577b1cd..a88c4652be35 100644 --- a/src/transformers/models/deprecated/transfo_xl/modeling_transfo_xl.py +++ b/src/transformers/models/deprecated/transfo_xl/modeling_transfo_xl.py @@ -876,10 +876,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, TransfoXLModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # the original code for Transformer-XL used shapes [len, bsz] but we want a unified interface in the library diff --git a/src/transformers/models/deprecated/tvlt/modeling_tvlt.py b/src/transformers/models/deprecated/tvlt/modeling_tvlt.py index d0083211bd30..6203215f11c9 100644 --- a/src/transformers/models/deprecated/tvlt/modeling_tvlt.py +++ b/src/transformers/models/deprecated/tvlt/modeling_tvlt.py @@ -725,10 +725,8 @@ def forward( >>> loss = outputs.loss ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict pixel_embedding_output, pixel_mask = self.pixel_embeddings(pixel_values, pixel_mask) diff --git a/src/transformers/models/deprecated/van/modeling_van.py b/src/transformers/models/deprecated/van/modeling_van.py index b31b0616ffa9..1a4ccb0bcd86 100644 --- a/src/transformers/models/deprecated/van/modeling_van.py +++ b/src/transformers/models/deprecated/van/modeling_van.py @@ -437,9 +437,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( diff --git a/src/transformers/models/deprecated/vit_hybrid/modeling_vit_hybrid.py b/src/transformers/models/deprecated/vit_hybrid/modeling_vit_hybrid.py index 553b0a7bb3bc..5b138db62a24 100644 --- a/src/transformers/models/deprecated/vit_hybrid/modeling_vit_hybrid.py +++ b/src/transformers/models/deprecated/vit_hybrid/modeling_vit_hybrid.py @@ -609,10 +609,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deprecated/xlm_prophetnet/modeling_xlm_prophetnet.py b/src/transformers/models/deprecated/xlm_prophetnet/modeling_xlm_prophetnet.py index e8b23b961e59..eda131c416aa 100644 --- a/src/transformers/models/deprecated/xlm_prophetnet/modeling_xlm_prophetnet.py +++ b/src/transformers/models/deprecated/xlm_prophetnet/modeling_xlm_prophetnet.py @@ -1280,10 +1280,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and inputs_embeds is None: @@ -1451,10 +1449,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ```""" use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and inputs_embeds is None: @@ -1803,10 +1799,8 @@ def forward( >>> last_hidden_states_ngram = outputs.last_hidden_state_ngram # predict hidden states ```""" use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if encoder_outputs is None: diff --git a/src/transformers/models/depth_anything/modeling_depth_anything.py b/src/transformers/models/depth_anything/modeling_depth_anything.py index 054eceb77d06..5693a059066f 100644 --- a/src/transformers/models/depth_anything/modeling_depth_anything.py +++ b/src/transformers/models/depth_anything/modeling_depth_anything.py @@ -390,10 +390,8 @@ def forward( raise NotImplementedError("Training is not implemented yet") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + outputs = self.backbone.forward_with_filtered_kwargs( pixel_values, output_hidden_states=output_hidden_states, output_attentions=output_attentions diff --git a/src/transformers/models/depth_pro/modeling_depth_pro.py b/src/transformers/models/depth_pro/modeling_depth_pro.py index d0bee293b0a6..cac1368a9d4a 100644 --- a/src/transformers/models/depth_pro/modeling_depth_pro.py +++ b/src/transformers/models/depth_pro/modeling_depth_pro.py @@ -698,10 +698,8 @@ def forward( >>> output.last_hidden_state.shape torch.Size([1, 35, 577, 1024]) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encodings = self.encoder( @@ -1107,10 +1105,8 @@ def forward( raise NotImplementedError("Training is not implemented yet") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + depth_pro_outputs = self.depth_pro( pixel_values=pixel_values, diff --git a/src/transformers/models/detr/modeling_detr.py b/src/transformers/models/detr/modeling_detr.py index d390c9a1a550..97467759730f 100644 --- a/src/transformers/models/detr/modeling_detr.py +++ b/src/transformers/models/detr/modeling_detr.py @@ -875,10 +875,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1003,10 +1001,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1189,10 +1185,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 100, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/dinat/modeling_dinat.py b/src/transformers/models/dinat/modeling_dinat.py index e598ffd8b152..384124b052fe 100644 --- a/src/transformers/models/dinat/modeling_dinat.py +++ b/src/transformers/models/dinat/modeling_dinat.py @@ -665,10 +665,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, DinatModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -853,10 +851,8 @@ def forward( [1, 512, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/dinov2/modeling_dinov2.py b/src/transformers/models/dinov2/modeling_dinov2.py index cace1ce316ad..bfdf29abdb7b 100644 --- a/src/transformers/models/dinov2/modeling_dinov2.py +++ b/src/transformers/models/dinov2/modeling_dinov2.py @@ -568,10 +568,8 @@ def forward( Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). Only relevant for pre-training. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -756,10 +754,8 @@ def forward( [1, 768, 16, 16] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/dinov2/modeling_flax_dinov2.py b/src/transformers/models/dinov2/modeling_flax_dinov2.py index b9ea2eaa3ebc..13d564b150d3 100644 --- a/src/transformers/models/dinov2/modeling_flax_dinov2.py +++ b/src/transformers/models/dinov2/modeling_flax_dinov2.py @@ -600,9 +600,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/dinov2_with_registers/modeling_dinov2_with_registers.py b/src/transformers/models/dinov2_with_registers/modeling_dinov2_with_registers.py index ba3ecf087487..2b3f18364e3f 100644 --- a/src/transformers/models/dinov2_with_registers/modeling_dinov2_with_registers.py +++ b/src/transformers/models/dinov2_with_registers/modeling_dinov2_with_registers.py @@ -587,10 +587,8 @@ def forward( Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). Only relevant for pre-training. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -778,10 +776,8 @@ def forward( [1, 768, 16, 16] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py b/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py index 48d425bfd4a0..474da26e9864 100644 --- a/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py +++ b/src/transformers/models/dinov2_with_registers/modular_dinov2_with_registers.py @@ -439,7 +439,7 @@ def forward( output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/distilbert/modeling_distilbert.py b/src/transformers/models/distilbert/modeling_distilbert.py index c5c1bdd0b3c0..c49600778a17 100755 --- a/src/transformers/models/distilbert/modeling_distilbert.py +++ b/src/transformers/models/distilbert/modeling_distilbert.py @@ -700,10 +700,8 @@ def forward( is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/distilbert/modeling_flax_distilbert.py b/src/transformers/models/distilbert/modeling_flax_distilbert.py index fba3dfd9d332..8c17e2a48b37 100644 --- a/src/transformers/models/distilbert/modeling_flax_distilbert.py +++ b/src/transformers/models/distilbert/modeling_flax_distilbert.py @@ -467,9 +467,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -510,9 +508,7 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict input_embeds = self.embeddings(input_ids, deterministic=deterministic) diff --git a/src/transformers/models/donut/modeling_donut_swin.py b/src/transformers/models/donut/modeling_donut_swin.py index 228606bca90f..bfb06ed9c2cf 100644 --- a/src/transformers/models/donut/modeling_donut_swin.py +++ b/src/transformers/models/donut/modeling_donut_swin.py @@ -944,10 +944,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/dpr/modeling_dpr.py b/src/transformers/models/dpr/modeling_dpr.py index 43415a5ed28a..045e25657d6b 100644 --- a/src/transformers/models/dpr/modeling_dpr.py +++ b/src/transformers/models/dpr/modeling_dpr.py @@ -366,10 +366,8 @@ def forward( >>> embeddings = model(input_ids).pooler_output ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: @@ -470,10 +468,8 @@ def forward( >>> embeddings = model(input_ids).pooler_output ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: @@ -574,10 +570,8 @@ def forward( >>> relevance_logits = outputs.relevance_logits ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/dpt/modeling_dpt.py b/src/transformers/models/dpt/modeling_dpt.py index 135a8d6ef3f8..c3df3869ed5a 100755 --- a/src/transformers/models/dpt/modeling_dpt.py +++ b/src/transformers/models/dpt/modeling_dpt.py @@ -890,10 +890,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndIntermediateActivations]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed @@ -1116,10 +1114,8 @@ def forward( raise NotImplementedError("Training is not implemented yet") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + if self.backbone is not None: outputs = self.backbone.forward_with_filtered_kwargs( @@ -1270,9 +1266,7 @@ def forward( >>> logits = outputs.logits ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and self.config.num_labels == 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/efficientnet/modeling_efficientnet.py b/src/transformers/models/efficientnet/modeling_efficientnet.py index 26410fc120a2..468c7c6c7a2f 100644 --- a/src/transformers/models/efficientnet/modeling_efficientnet.py +++ b/src/transformers/models/efficientnet/modeling_efficientnet.py @@ -476,9 +476,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/electra/modeling_electra.py b/src/transformers/models/electra/modeling_electra.py index eab5c1a9fd50..9248a538e1b2 100644 --- a/src/transformers/models/electra/modeling_electra.py +++ b/src/transformers/models/electra/modeling_electra.py @@ -750,10 +750,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/electra/modeling_flax_electra.py b/src/transformers/models/electra/modeling_flax_electra.py index 14d845476d62..23b2e262471a 100644 --- a/src/transformers/models/electra/modeling_flax_electra.py +++ b/src/transformers/models/electra/modeling_flax_electra.py @@ -786,9 +786,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/emu3/modeling_emu3.py b/src/transformers/models/emu3/modeling_emu3.py index 0e39ddf243c5..23886841930b 100644 --- a/src/transformers/models/emu3/modeling_emu3.py +++ b/src/transformers/models/emu3/modeling_emu3.py @@ -1478,10 +1478,8 @@ def forward( [`AutoImageProcessor`]. See [`Emu3ImageProcessor.__call__`] for details ([]`Emu3Processor`] uses [`Emu3ImageProcessor`] for processing images). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): @@ -1634,10 +1632,8 @@ def forward( >>> generated_ids = model.generate(**inputs, max_new_tokens=100, do_sample=False) >>> processor.batch_decode(generated_ids, skip_special_tokens=True)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/emu3/modular_emu3.py b/src/transformers/models/emu3/modular_emu3.py index c1b285317fb3..a877322a36fb 100644 --- a/src/transformers/models/emu3/modular_emu3.py +++ b/src/transformers/models/emu3/modular_emu3.py @@ -1015,7 +1015,7 @@ def forward( [`AutoImageProcessor`]. See [`Emu3ImageProcessor.__call__`] for details ([]`Emu3Processor`] uses [`Emu3ImageProcessor`] for processing images). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1170,7 +1170,7 @@ def forward( >>> generated_ids = model.generate(**inputs, max_new_tokens=100, do_sample=False) >>> processor.batch_decode(generated_ids, skip_special_tokens=True)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py b/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py index 13da12b1aa76..2d4bf797b070 100644 --- a/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py +++ b/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py @@ -457,9 +457,7 @@ def encode( >>> encoder_outputs = model.encode(input_ids) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -541,9 +539,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -674,9 +670,7 @@ def __call__( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs diff --git a/src/transformers/models/ernie/modeling_ernie.py b/src/transformers/models/ernie/modeling_ernie.py index 18cf43b5f2d3..060351b64876 100644 --- a/src/transformers/models/ernie/modeling_ernie.py +++ b/src/transformers/models/ernie/modeling_ernie.py @@ -762,10 +762,8 @@ def forward( assign a `task_type_id` to each task and the `task_type_id` is in the range `[0, config.task_type_vocab_size-1] """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/esm/modeling_esm.py b/src/transformers/models/esm/modeling_esm.py index cc0c8f761706..c19c053ecb76 100755 --- a/src/transformers/models/esm/modeling_esm.py +++ b/src/transformers/models/esm/modeling_esm.py @@ -902,10 +902,8 @@ def forward( is useful if you want more control over how to convert `input_ids` indices into associated vectors than the model's internal embedding lookup matrix. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/falcon/modeling_falcon.py b/src/transformers/models/falcon/modeling_falcon.py index f7a6d9398307..f03502ef49d5 100644 --- a/src/transformers/models/falcon/modeling_falcon.py +++ b/src/transformers/models/falcon/modeling_falcon.py @@ -752,10 +752,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/falcon_h1/modeling_falcon_h1.py b/src/transformers/models/falcon_h1/modeling_falcon_h1.py index 662d6254c841..ab8d0d675227 100644 --- a/src/transformers/models/falcon_h1/modeling_falcon_h1.py +++ b/src/transformers/models/falcon_h1/modeling_falcon_h1.py @@ -1269,10 +1269,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1551,10 +1549,8 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs = self.model( diff --git a/src/transformers/models/falcon_h1/modular_falcon_h1.py b/src/transformers/models/falcon_h1/modular_falcon_h1.py index 14a89f2afe95..96e15f5cd162 100644 --- a/src/transformers/models/falcon_h1/modular_falcon_h1.py +++ b/src/transformers/models/falcon_h1/modular_falcon_h1.py @@ -1043,7 +1043,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwargs, for now ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1285,7 +1285,7 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/falcon_mamba/modeling_falcon_mamba.py b/src/transformers/models/falcon_mamba/modeling_falcon_mamba.py index 257f2c50cd93..1f149d258bde 100644 --- a/src/transformers/models/falcon_mamba/modeling_falcon_mamba.py +++ b/src/transformers/models/falcon_mamba/modeling_falcon_mamba.py @@ -585,9 +585,7 @@ def forward( use_cache (`bool`, *optional*): If set to `True`, the `cache_params` is returned and can be used to quickly generate the next logits. """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else (self.config.use_cache if not self.training else False) return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/fastspeech2_conformer/modeling_fastspeech2_conformer.py b/src/transformers/models/fastspeech2_conformer/modeling_fastspeech2_conformer.py index 60757a113a5b..cb4be41fc35d 100644 --- a/src/transformers/models/fastspeech2_conformer/modeling_fastspeech2_conformer.py +++ b/src/transformers/models/fastspeech2_conformer/modeling_fastspeech2_conformer.py @@ -1176,10 +1176,8 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if attention_mask is None: attention_mask = torch.ones(input_ids.shape, device=input_ids.device) @@ -1571,9 +1569,7 @@ def forward( output_attentions = ( output_attentions if output_attentions is not None else self.config.model_config.output_attentions ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.model_config.output_hidden_states - ) + model_outputs = self.model( input_ids, diff --git a/src/transformers/models/flaubert/modeling_flaubert.py b/src/transformers/models/flaubert/modeling_flaubert.py index f1019480b56d..2b4ddd91cf6f 100644 --- a/src/transformers/models/flaubert/modeling_flaubert.py +++ b/src/transformers/models/flaubert/modeling_flaubert.py @@ -832,10 +832,8 @@ def forward( See usage examples detailed in the [multilingual documentation](../multilingual). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # removed: src_enc=None, src_len=None diff --git a/src/transformers/models/flava/modeling_flava.py b/src/transformers/models/flava/modeling_flava.py index e372f521bf23..f51bbf83462e 100644 --- a/src/transformers/models/flava/modeling_flava.py +++ b/src/transformers/models/flava/modeling_flava.py @@ -777,10 +777,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, image_num_patches)`): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -879,10 +877,8 @@ def forward( - 1 corresponds to a *sentence B* token. [What are token type IDs?](../glossary#token-type-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None: @@ -978,10 +974,8 @@ def forward( hidden_states (`torch.FloatTensor` of shape `(batch_size, image_num_patches + text_seq_len, hidden_size)`): The concatenated hidden states of unimodal encoders. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, seq_length, _ = hidden_states.size() diff --git a/src/transformers/models/fnet/modeling_fnet.py b/src/transformers/models/fnet/modeling_fnet.py index 619d6c9c5add..77e493570a34 100755 --- a/src/transformers/models/fnet/modeling_fnet.py +++ b/src/transformers/models/fnet/modeling_fnet.py @@ -477,9 +477,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/focalnet/modeling_focalnet.py b/src/transformers/models/focalnet/modeling_focalnet.py index 54abac6625b2..57b1bf9ea401 100644 --- a/src/transformers/models/focalnet/modeling_focalnet.py +++ b/src/transformers/models/focalnet/modeling_focalnet.py @@ -669,9 +669,7 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -944,9 +942,7 @@ def forward( >>> outputs = model(**inputs) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + outputs = self.focalnet(pixel_values, output_hidden_states=True, return_dict=True) diff --git a/src/transformers/models/fsmt/modeling_fsmt.py b/src/transformers/models/fsmt/modeling_fsmt.py index d3830b85b216..6ba0fb6654fe 100644 --- a/src/transformers/models/fsmt/modeling_fsmt.py +++ b/src/transformers/models/fsmt/modeling_fsmt.py @@ -977,10 +977,8 @@ def forward( if decoder_input_ids is None: use_cache = False - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/funnel/modeling_funnel.py b/src/transformers/models/funnel/modeling_funnel.py index a134ab9c7f13..8fa8b881f4ae 100644 --- a/src/transformers/models/funnel/modeling_funnel.py +++ b/src/transformers/models/funnel/modeling_funnel.py @@ -866,10 +866,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: @@ -932,10 +930,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/fuyu/modeling_fuyu.py b/src/transformers/models/fuyu/modeling_fuyu.py index c4ec5e829288..915573d721e5 100644 --- a/src/transformers/models/fuyu/modeling_fuyu.py +++ b/src/transformers/models/fuyu/modeling_fuyu.py @@ -167,10 +167,8 @@ def forward( image_patches_indices (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): Tensor of indices of the image patches in the input_ids tensor. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -314,10 +312,8 @@ def forward( A blue bus parked on the side of a road. ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/gemma/modeling_flax_gemma.py b/src/transformers/models/gemma/modeling_flax_gemma.py index 0addcd7dde7a..bb87a936ddfd 100644 --- a/src/transformers/models/gemma/modeling_flax_gemma.py +++ b/src/transformers/models/gemma/modeling_flax_gemma.py @@ -494,9 +494,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/gemma/modeling_gemma.py b/src/transformers/models/gemma/modeling_gemma.py index 4c6e1f848602..e8f0ad18ab47 100644 --- a/src/transformers/models/gemma/modeling_gemma.py +++ b/src/transformers/models/gemma/modeling_gemma.py @@ -375,10 +375,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/gemma/modular_gemma.py b/src/transformers/models/gemma/modular_gemma.py index 5532dbaf7a7f..f2e217272c39 100644 --- a/src/transformers/models/gemma/modular_gemma.py +++ b/src/transformers/models/gemma/modular_gemma.py @@ -381,7 +381,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/gemma2/modeling_gemma2.py b/src/transformers/models/gemma2/modeling_gemma2.py index bda7b81b3d43..9f66bf63b569 100644 --- a/src/transformers/models/gemma2/modeling_gemma2.py +++ b/src/transformers/models/gemma2/modeling_gemma2.py @@ -401,10 +401,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -588,10 +586,8 @@ def forward( "It is strongly recommended to train Gemma2 models with the `eager` attention implementation " f"instead of `{self.config._attn_implementation}`. Use `eager` with `AutoModelForCausalLM.from_pretrained('', attn_implementation='eager')`." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, diff --git a/src/transformers/models/gemma2/modular_gemma2.py b/src/transformers/models/gemma2/modular_gemma2.py index 2a2e43bfe0b2..3c964ea728e5 100644 --- a/src/transformers/models/gemma2/modular_gemma2.py +++ b/src/transformers/models/gemma2/modular_gemma2.py @@ -382,7 +382,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -534,7 +534,7 @@ def forward( "It is strongly recommended to train Gemma2 models with the `eager` attention implementation " f"instead of `{self.config._attn_implementation}`. Use `eager` with `AutoModelForCausalLM.from_pretrained('', attn_implementation='eager')`." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/gemma3/modeling_gemma3.py b/src/transformers/models/gemma3/modeling_gemma3.py index c42504e8ba1e..dd30fb76c9e6 100644 --- a/src/transformers/models/gemma3/modeling_gemma3.py +++ b/src/transformers/models/gemma3/modeling_gemma3.py @@ -519,10 +519,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -707,10 +705,8 @@ def forward( "It is strongly recommended to train Gemma3 models with the `eager` attention implementation " f"instead of `{self.config._attn_implementation}`. Use `eager` with `AutoModelForCausalLM.from_pretrained('', attn_implementation='eager')`." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( input_ids=input_ids, @@ -893,10 +889,8 @@ def forward( if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Replace image id woth PAD if the image token if OOV, to avoid index-errors @@ -1100,10 +1094,8 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/gemma3/modular_gemma3.py b/src/transformers/models/gemma3/modular_gemma3.py index ac076fd26037..420e64ad5047 100644 --- a/src/transformers/models/gemma3/modular_gemma3.py +++ b/src/transformers/models/gemma3/modular_gemma3.py @@ -568,7 +568,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -781,7 +781,7 @@ def forward( if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -934,7 +934,7 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/git/modeling_git.py b/src/transformers/models/git/modeling_git.py index 6068ce169da3..603eabcb6545 100644 --- a/src/transformers/models/git/modeling_git.py +++ b/src/transformers/models/git/modeling_git.py @@ -827,10 +827,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -892,10 +890,8 @@ def forward( interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1128,10 +1124,8 @@ def forward( >>> outputs = model(**inputs) >>> last_hidden_state = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/glpn/modeling_glpn.py b/src/transformers/models/glpn/modeling_glpn.py index 299eef53d1c6..328f61301013 100755 --- a/src/transformers/models/glpn/modeling_glpn.py +++ b/src/transformers/models/glpn/modeling_glpn.py @@ -454,10 +454,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( @@ -678,9 +676,7 @@ def forward( >>> depth = Image.fromarray(depth.astype("uint8")) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + outputs = self.glpn( pixel_values, diff --git a/src/transformers/models/got_ocr2/modeling_got_ocr2.py b/src/transformers/models/got_ocr2/modeling_got_ocr2.py index bcf30f585250..d490980fb1dc 100644 --- a/src/transformers/models/got_ocr2/modeling_got_ocr2.py +++ b/src/transformers/models/got_ocr2/modeling_got_ocr2.py @@ -444,10 +444,8 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, ) -> GotOcr2VisionEncoderOutput: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -669,10 +667,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, GotOcr2ModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): @@ -842,10 +838,8 @@ def forward( "You should keep in mind what features from the module should be used, especially when you're planning to sell a template." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/got_ocr2/modular_got_ocr2.py b/src/transformers/models/got_ocr2/modular_got_ocr2.py index 32ef0abb9ce3..f53cff08b14f 100644 --- a/src/transformers/models/got_ocr2/modular_got_ocr2.py +++ b/src/transformers/models/got_ocr2/modular_got_ocr2.py @@ -332,7 +332,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, GotOcr2ModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -437,7 +437,7 @@ def forward( "You should keep in mind what features from the module should be used, especially when you're planning to sell a template." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/gpt2/modeling_flax_gpt2.py b/src/transformers/models/gpt2/modeling_flax_gpt2.py index 8e419217c5a3..d22ff5d63f36 100644 --- a/src/transformers/models/gpt2/modeling_flax_gpt2.py +++ b/src/transformers/models/gpt2/modeling_flax_gpt2.py @@ -470,9 +470,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/gpt2/modeling_gpt2.py b/src/transformers/models/gpt2/modeling_gpt2.py index 30012b583748..9f1fc4f2aa19 100644 --- a/src/transformers/models/gpt2/modeling_gpt2.py +++ b/src/transformers/models/gpt2/modeling_gpt2.py @@ -803,10 +803,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py b/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py index 297c30cb06a1..b11f6513c3d9 100644 --- a/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py +++ b/src/transformers/models/gpt_bigcode/modeling_gpt_bigcode.py @@ -747,10 +747,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py b/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py index a6cdc50b359b..d92eef20a785 100644 --- a/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py +++ b/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py @@ -413,9 +413,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/gpt_neo/modeling_gpt_neo.py b/src/transformers/models/gpt_neo/modeling_gpt_neo.py index c62cb3aaf524..7e7227a85158 100755 --- a/src/transformers/models/gpt_neo/modeling_gpt_neo.py +++ b/src/transformers/models/gpt_neo/modeling_gpt_neo.py @@ -560,10 +560,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/gpt_neox/modeling_gpt_neox.py b/src/transformers/models/gpt_neox/modeling_gpt_neox.py index 8e00022b0b7b..1d64b2624128 100755 --- a/src/transformers/models/gpt_neox/modeling_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modeling_gpt_neox.py @@ -354,10 +354,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/gpt_neox/modular_gpt_neox.py b/src/transformers/models/gpt_neox/modular_gpt_neox.py index 58af6ff0f2fd..37ce43883163 100644 --- a/src/transformers/models/gpt_neox/modular_gpt_neox.py +++ b/src/transformers/models/gpt_neox/modular_gpt_neox.py @@ -299,7 +299,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/gpt_neox_japanese/modeling_gpt_neox_japanese.py b/src/transformers/models/gpt_neox_japanese/modeling_gpt_neox_japanese.py index dbcb831ea218..2cf4830a1791 100755 --- a/src/transformers/models/gpt_neox_japanese/modeling_gpt_neox_japanese.py +++ b/src/transformers/models/gpt_neox_japanese/modeling_gpt_neox_japanese.py @@ -447,10 +447,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict use_cache = use_cache if use_cache is not None else self.config.use_cache diff --git a/src/transformers/models/gptj/modeling_flax_gptj.py b/src/transformers/models/gptj/modeling_flax_gptj.py index 12ea7a4fffb4..42c7c2c4338c 100644 --- a/src/transformers/models/gptj/modeling_flax_gptj.py +++ b/src/transformers/models/gptj/modeling_flax_gptj.py @@ -447,9 +447,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/gptj/modeling_gptj.py b/src/transformers/models/gptj/modeling_gptj.py index 59871c48377c..8ed7b5201608 100644 --- a/src/transformers/models/gptj/modeling_gptj.py +++ b/src/transformers/models/gptj/modeling_gptj.py @@ -648,10 +648,8 @@ def forward( is useful if you want more control over how to convert *input_ids* indices into associated vectors than the model's internal embedding lookup matrix. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/granite/modeling_granite.py b/src/transformers/models/granite/modeling_granite.py index 634032d7a236..812dec5d67f8 100644 --- a/src/transformers/models/granite/modeling_granite.py +++ b/src/transformers/models/granite/modeling_granite.py @@ -402,10 +402,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -561,10 +559,8 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/granite/modular_granite.py b/src/transformers/models/granite/modular_granite.py index 6ab58b1c67f6..be196731508e 100644 --- a/src/transformers/models/granite/modular_granite.py +++ b/src/transformers/models/granite/modular_granite.py @@ -142,7 +142,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -244,7 +244,7 @@ def forward( logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs: Unpack[KwargsForCausalLM], ) -> CausalLMOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/granite_speech/modeling_granite_speech.py b/src/transformers/models/granite_speech/modeling_granite_speech.py index 377904accbf3..597d9e815f3e 100644 --- a/src/transformers/models/granite_speech/modeling_granite_speech.py +++ b/src/transformers/models/granite_speech/modeling_granite_speech.py @@ -390,10 +390,8 @@ def forward( Mask to be applied to audio features prior to scattering into the language embeddings. """ # TODO (@alex-jw-brooks) add an example to this docstring once models are released - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/granitemoe/modeling_granitemoe.py b/src/transformers/models/granitemoe/modeling_granitemoe.py index ffaf86697a60..08db5f5a711a 100644 --- a/src/transformers/models/granitemoe/modeling_granitemoe.py +++ b/src/transformers/models/granitemoe/modeling_granitemoe.py @@ -665,10 +665,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -968,13 +966,11 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/granitemoehybrid/modeling_granitemoehybrid.py b/src/transformers/models/granitemoehybrid/modeling_granitemoehybrid.py index 51785d48f4cc..47425b5e88a1 100644 --- a/src/transformers/models/granitemoehybrid/modeling_granitemoehybrid.py +++ b/src/transformers/models/granitemoehybrid/modeling_granitemoehybrid.py @@ -1282,10 +1282,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1677,13 +1675,11 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py b/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py index 4721dd673fbb..aacad8de2ff6 100644 --- a/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py +++ b/src/transformers/models/granitemoehybrid/modular_granitemoehybrid.py @@ -206,7 +206,7 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/granitemoeshared/modeling_granitemoeshared.py b/src/transformers/models/granitemoeshared/modeling_granitemoeshared.py index 8cd13f8a8c55..289b25a76bd6 100644 --- a/src/transformers/models/granitemoeshared/modeling_granitemoeshared.py +++ b/src/transformers/models/granitemoeshared/modeling_granitemoeshared.py @@ -613,10 +613,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -996,13 +994,11 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/grounding_dino/modeling_grounding_dino.py b/src/transformers/models/grounding_dino/modeling_grounding_dino.py index 9603651fafcd..4322e0f250fe 100644 --- a/src/transformers/models/grounding_dino/modeling_grounding_dino.py +++ b/src/transformers/models/grounding_dino/modeling_grounding_dino.py @@ -1583,10 +1583,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict reference_points = self.get_reference_points(spatial_shapes, valid_ratios, device=vision_features.device) @@ -1733,10 +1731,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -2125,10 +2121,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 900, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_self_attention_masks, position_ids = generate_masks_with_special_tokens_and_transfer_map(input_ids) diff --git a/src/transformers/models/groupvit/modeling_groupvit.py b/src/transformers/models/groupvit/modeling_groupvit.py index d5e37cdc7f77..6f0a83561e08 100644 --- a/src/transformers/models/groupvit/modeling_groupvit.py +++ b/src/transformers/models/groupvit/modeling_groupvit.py @@ -564,7 +564,7 @@ def __init__( self.activation_fn = ACT2FN[config.hidden_act] hidden_size = hidden_size if hidden_size is not None else config.hidden_size intermediate_size = intermediate_size if intermediate_size is not None else config.intermediate_size - output_size = output_size if output_size is not None else hidden_size + self.fc1 = nn.Linear(hidden_size, intermediate_size) self.fc2 = nn.Linear(intermediate_size, output_size) @@ -807,10 +807,8 @@ def forward( output_attentions: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_hidden_states = () if output_hidden_states else None @@ -892,10 +890,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -957,10 +953,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None: @@ -1094,10 +1088,8 @@ def forward( output_attentions: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1250,10 +1242,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use GROUPVIT model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1300,10 +1290,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use GROUPVIT model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1358,15 +1346,13 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use GROUPVIT model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_segmentation = ( output_segmentation if output_segmentation is not None else self.config.output_segmentation ) if output_segmentation: output_attentions = True - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/hgnet_v2/modeling_hgnet_v2.py b/src/transformers/models/hgnet_v2/modeling_hgnet_v2.py index 4411a5048812..7d028f739b30 100644 --- a/src/transformers/models/hgnet_v2/modeling_hgnet_v2.py +++ b/src/transformers/models/hgnet_v2/modeling_hgnet_v2.py @@ -377,9 +377,7 @@ def forward( [1, 2048, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embedder(pixel_values) @@ -461,9 +459,7 @@ def forward( torch.Size([1, 2]) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embedder(pixel_values) outputs = self.encoder(embedding_output, output_hidden_states=output_hidden_states, return_dict=return_dict) last_hidden_state = outputs[0] diff --git a/src/transformers/models/hiera/modeling_hiera.py b/src/transformers/models/hiera/modeling_hiera.py index e542c7b69a55..617c5b827286 100644 --- a/src/transformers/models/hiera/modeling_hiera.py +++ b/src/transformers/models/hiera/modeling_hiera.py @@ -910,10 +910,8 @@ def forward( noise (`torch.FloatTensor` of shape `(batch_size, num_mask_units)`, *optional*): Mainly used for testing purposes to control randomness and maintain the reproducibility """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1229,10 +1227,8 @@ def forward( [1, 196, 768] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs = self.hiera( pixel_values, @@ -1330,10 +1326,8 @@ def forward( `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs = self.hiera( pixel_values, @@ -1447,10 +1441,8 @@ def forward( [1, 768, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output, _, _ = self.embeddings(pixel_values) diff --git a/src/transformers/models/hubert/modeling_hubert.py b/src/transformers/models/hubert/modeling_hubert.py index faa0ff48c688..9ef8b36cc273 100755 --- a/src/transformers/models/hubert/modeling_hubert.py +++ b/src/transformers/models/hubert/modeling_hubert.py @@ -1024,10 +1024,8 @@ def forward( >>> input_values = processor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1 >>> hidden_states = model(input_values).last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/hubert/modular_hubert.py b/src/transformers/models/hubert/modular_hubert.py index 9de202cd6dd9..e8501831557b 100644 --- a/src/transformers/models/hubert/modular_hubert.py +++ b/src/transformers/models/hubert/modular_hubert.py @@ -256,7 +256,7 @@ def forward( >>> input_values = processor(ds["speech"][0], return_tensors="pt").input_values # Batch size 1 >>> hidden_states = model(input_values).last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/ibert/modeling_ibert.py b/src/transformers/models/ibert/modeling_ibert.py index 89fd716f885f..beace48c7e66 100644 --- a/src/transformers/models/ibert/modeling_ibert.py +++ b/src/transformers/models/ibert/modeling_ibert.py @@ -706,10 +706,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPoolingAndCrossAttentions, tuple[torch.FloatTensor]]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/idefics/modeling_idefics.py b/src/transformers/models/idefics/modeling_idefics.py index f4a6715d6cf0..9fe4a67640e0 100644 --- a/src/transformers/models/idefics/modeling_idefics.py +++ b/src/transformers/models/idefics/modeling_idefics.py @@ -1069,10 +1069,8 @@ def forward( """ device = input_ids.device if input_ids is not None else inputs_embeds.device - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1559,10 +1557,8 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/idefics/modeling_tf_idefics.py b/src/transformers/models/idefics/modeling_tf_idefics.py index 662fe4d88d13..c6309811ed35 100644 --- a/src/transformers/models/idefics/modeling_tf_idefics.py +++ b/src/transformers/models/idefics/modeling_tf_idefics.py @@ -1282,9 +1282,7 @@ def call( training: Optional[bool] = None, ) -> Union[TFIdeficsBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1712,9 +1710,7 @@ def call( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/idefics/vision.py b/src/transformers/models/idefics/vision.py index 815b902d3fba..729466c3f1f4 100644 --- a/src/transformers/models/idefics/vision.py +++ b/src/transformers/models/idefics/vision.py @@ -387,7 +387,7 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -456,7 +456,7 @@ def forward( Returns: """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/idefics/vision_tf.py b/src/transformers/models/idefics/vision_tf.py index 9db18f265db5..67f1208dba9b 100644 --- a/src/transformers/models/idefics/vision_tf.py +++ b/src/transformers/models/idefics/vision_tf.py @@ -437,7 +437,7 @@ def call( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -521,7 +521,7 @@ def call( Returns: """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/idefics2/modeling_idefics2.py b/src/transformers/models/idefics2/modeling_idefics2.py index d9b5d5e68336..e3f5253d2d1f 100644 --- a/src/transformers/models/idefics2/modeling_idefics2.py +++ b/src/transformers/models/idefics2/modeling_idefics2.py @@ -435,10 +435,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -554,10 +552,8 @@ def forward( patch_attention_mask (`torch.BoolTensor` of shape `(batch_size, num_patches_height, num_patches_width)`, *optional*): The attention mask for the patches. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size = pixel_values.size(0) @@ -1048,10 +1044,8 @@ def forward( image_hidden_states (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`): The hidden states of the image encoder after modality projection and perceiver resampling. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1259,10 +1253,8 @@ def forward( ['In this image, we can see the city of New York, and more specifically the Statue of Liberty. In this image, we can see the city of New York, and more specifically the Statue of Liberty.\n\n', 'In which city is that bridge located?\n\nThe bridge is located in the city of Pittsburgh, Pennsylvania.\n\n\nThe bridge is'] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/idefics3/modeling_idefics3.py b/src/transformers/models/idefics3/modeling_idefics3.py index 53b3cc2e304e..eafad7b2d0a0 100644 --- a/src/transformers/models/idefics3/modeling_idefics3.py +++ b/src/transformers/models/idefics3/modeling_idefics3.py @@ -396,10 +396,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -563,10 +561,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size = pixel_values.size(0) @@ -781,10 +777,8 @@ def forward( image_hidden_states (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`): The hidden states of the image encoder after modality projection. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -997,10 +991,8 @@ def forward( >>> print(generated_texts[1]) Assistant: The bridge is in San Francisco. ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/ijepa/modeling_ijepa.py b/src/transformers/models/ijepa/modeling_ijepa.py index c7ce7c29b4cf..22febfa65a83 100644 --- a/src/transformers/models/ijepa/modeling_ijepa.py +++ b/src/transformers/models/ijepa/modeling_ijepa.py @@ -511,10 +511,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/imagegpt/modeling_imagegpt.py b/src/transformers/models/imagegpt/modeling_imagegpt.py index ebcec593fe15..5f1744b76932 100755 --- a/src/transformers/models/imagegpt/modeling_imagegpt.py +++ b/src/transformers/models/imagegpt/modeling_imagegpt.py @@ -610,10 +610,8 @@ def forward( input_ids = kwargs.pop("pixel_values") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/informer/modeling_informer.py b/src/transformers/models/informer/modeling_informer.py index 6f59e0ee14bd..d3f3c533ca33 100644 --- a/src/transformers/models/informer/modeling_informer.py +++ b/src/transformers/models/informer/modeling_informer.py @@ -1045,10 +1045,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.value_embedding(inputs_embeds) @@ -1225,10 +1223,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1645,10 +1641,8 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/informer/modular_informer.py b/src/transformers/models/informer/modular_informer.py index db3d52564d5e..2d01a9d1b1fd 100644 --- a/src/transformers/models/informer/modular_informer.py +++ b/src/transformers/models/informer/modular_informer.py @@ -568,7 +568,7 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/instructblip/modeling_instructblip.py b/src/transformers/models/instructblip/modeling_instructblip.py index 1708a86082bc..f8f945f7d4e1 100644 --- a/src/transformers/models/instructblip/modeling_instructblip.py +++ b/src/transformers/models/instructblip/modeling_instructblip.py @@ -411,10 +411,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -471,10 +469,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1091,10 +1087,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and query_embeds is None: diff --git a/src/transformers/models/instructblipvideo/modeling_instructblipvideo.py b/src/transformers/models/instructblipvideo/modeling_instructblipvideo.py index 8c0d5c05a3b0..be908e9c0659 100644 --- a/src/transformers/models/instructblipvideo/modeling_instructblipvideo.py +++ b/src/transformers/models/instructblipvideo/modeling_instructblipvideo.py @@ -340,10 +340,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -907,10 +905,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1057,10 +1053,8 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and query_embeds is None: diff --git a/src/transformers/models/internvl/modeling_internvl.py b/src/transformers/models/internvl/modeling_internvl.py index 3958c780182f..9582a6ce9c5a 100644 --- a/src/transformers/models/internvl/modeling_internvl.py +++ b/src/transformers/models/internvl/modeling_internvl.py @@ -506,10 +506,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + embedding_output, _ = self.embeddings(pixel_values, bool_masked_pos=bool_masked_pos) @@ -699,10 +697,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, InternVLModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -972,10 +968,8 @@ def forward( >>> print(processor.decode(generate_ids[0, inputs["input_ids"].shape[1] :], skip_special_tokens=True)) The images depict the Statue of Liberty and the Golden Gate Bridge. ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/internvl/modular_internvl.py b/src/transformers/models/internvl/modular_internvl.py index 2c567e28d86e..1c7e02d5526c 100644 --- a/src/transformers/models/internvl/modular_internvl.py +++ b/src/transformers/models/internvl/modular_internvl.py @@ -456,7 +456,7 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -624,7 +624,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, InternVLModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/jamba/modeling_jamba.py b/src/transformers/models/jamba/modeling_jamba.py index 52570d8f7f8e..ad297d2b00de 100755 --- a/src/transformers/models/jamba/modeling_jamba.py +++ b/src/transformers/models/jamba/modeling_jamba.py @@ -1140,13 +1140,11 @@ def forward( output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1350,14 +1348,12 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/janus/modeling_janus.py b/src/transformers/models/janus/modeling_janus.py index 1017bc1c118c..ae28f671b58a 100644 --- a/src/transformers/models/janus/modeling_janus.py +++ b/src/transformers/models/janus/modeling_janus.py @@ -495,10 +495,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -554,10 +552,8 @@ def forward( return_dict: Optional[bool] = None, interpolate_pos_encoding: bool = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1100,10 +1096,8 @@ def forward( logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError( @@ -1219,10 +1213,8 @@ def forward( config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs = self.model( input_ids=input_ids, diff --git a/src/transformers/models/janus/modular_janus.py b/src/transformers/models/janus/modular_janus.py index 4814dc4db239..847e9cc7d8fd 100644 --- a/src/transformers/models/janus/modular_janus.py +++ b/src/transformers/models/janus/modular_janus.py @@ -936,7 +936,7 @@ def forward( logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1054,7 +1054,7 @@ def forward( config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/jetmoe/modeling_jetmoe.py b/src/transformers/models/jetmoe/modeling_jetmoe.py index 5753c616553c..58ab6e20269a 100644 --- a/src/transformers/models/jetmoe/modeling_jetmoe.py +++ b/src/transformers/models/jetmoe/modeling_jetmoe.py @@ -894,10 +894,8 @@ def forward( output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) @@ -1203,10 +1201,8 @@ def forward( (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/kosmos2/modeling_kosmos2.py b/src/transformers/models/kosmos2/modeling_kosmos2.py index 34dc0848b79c..195110c32b92 100644 --- a/src/transformers/models/kosmos2/modeling_kosmos2.py +++ b/src/transformers/models/kosmos2/modeling_kosmos2.py @@ -508,10 +508,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -573,10 +571,8 @@ def forward( interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1060,10 +1056,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1716,10 +1710,8 @@ def forward( >>> list(last_hidden_state.shape) [1, 91, 2048] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_model_output = None @@ -1863,10 +1855,8 @@ def forward( >>> entities [('a snowman', (12, 21), [(0.390625, 0.046875, 0.984375, 0.828125)]), ('a fire', (41, 47), [(0.171875, 0.015625, 0.484375, 0.890625)])] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_model_output = None projection_attentions = None diff --git a/src/transformers/models/layoutlm/modeling_layoutlm.py b/src/transformers/models/layoutlm/modeling_layoutlm.py index 81eb5aaa94c3..1505acc7684d 100644 --- a/src/transformers/models/layoutlm/modeling_layoutlm.py +++ b/src/transformers/models/layoutlm/modeling_layoutlm.py @@ -712,10 +712,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py b/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py index fb91e3c8e3a4..1dafbedb844b 100755 --- a/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py +++ b/src/transformers/models/layoutlmv2/modeling_layoutlmv2.py @@ -775,10 +775,8 @@ def forward( torch.Size([1, 342, 768]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict input_shape = self._get_input_shape(input_ids, inputs_embeds) diff --git a/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py b/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py index c6873d957e77..62d87290c883 100644 --- a/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py +++ b/src/transformers/models/layoutlmv3/modeling_layoutlmv3.py @@ -756,10 +756,8 @@ def forward( >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None: diff --git a/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py b/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py index d6703420e4ad..87cdeb317fd1 100644 --- a/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py +++ b/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py @@ -979,9 +979,7 @@ def call( # The complexity of this method is mostly just due to handling of these different modalities. output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if input_ids is not None: diff --git a/src/transformers/models/led/modeling_led.py b/src/transformers/models/led/modeling_led.py index c840e85ead88..bb9261c986ae 100755 --- a/src/transformers/models/led/modeling_led.py +++ b/src/transformers/models/led/modeling_led.py @@ -1606,10 +1606,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # check input_ids and inputs_embeds @@ -1858,10 +1856,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2084,10 +2080,8 @@ def forward( - 0 for local attention (a sliding window attention), - 1 for global attention (tokens that attend to all other tokens, and all other tokens attend to them). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/levit/modeling_levit.py b/src/transformers/models/levit/modeling_levit.py index 0e51a4ff1d6a..8c60276b747d 100644 --- a/src/transformers/models/levit/modeling_levit.py +++ b/src/transformers/models/levit/modeling_levit.py @@ -505,9 +505,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/lightglue/modeling_lightglue.py b/src/transformers/models/lightglue/modeling_lightglue.py index 4df4888621e1..2ba19b89bce2 100644 --- a/src/transformers/models/lightglue/modeling_lightglue.py +++ b/src/transformers/models/lightglue/modeling_lightglue.py @@ -880,10 +880,8 @@ def forward( if labels is not None: raise ValueError("LightGlue is not trainable, no labels should be provided.") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values.ndim != 5 or pixel_values.size(1) != 2: raise ValueError("Input must be a 5D tensor of shape (batch_size, 2, num_channels, height, width)") diff --git a/src/transformers/models/lightglue/modular_lightglue.py b/src/transformers/models/lightglue/modular_lightglue.py index 96a389194b46..be1e50883e52 100644 --- a/src/transformers/models/lightglue/modular_lightglue.py +++ b/src/transformers/models/lightglue/modular_lightglue.py @@ -954,7 +954,7 @@ def forward( if labels is not None: raise ValueError("LightGlue is not trainable, no labels should be provided.") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/lilt/modeling_lilt.py b/src/transformers/models/lilt/modeling_lilt.py index 24a35ccfce1b..2a0756debc85 100644 --- a/src/transformers/models/lilt/modeling_lilt.py +++ b/src/transformers/models/lilt/modeling_lilt.py @@ -662,10 +662,8 @@ def forward( >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/llama/modeling_flax_llama.py b/src/transformers/models/llama/modeling_flax_llama.py index 63e34e996ade..2b24b5218933 100644 --- a/src/transformers/models/llama/modeling_flax_llama.py +++ b/src/transformers/models/llama/modeling_flax_llama.py @@ -476,9 +476,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/llama4/modeling_llama4.py b/src/transformers/models/llama4/modeling_llama4.py index 9cdcc07ad960..43897f6ea59b 100644 --- a/src/transformers/models/llama4/modeling_llama4.py +++ b/src/transformers/models/llama4/modeling_llama4.py @@ -511,10 +511,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -693,10 +691,8 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) @@ -1021,10 +1017,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -1173,10 +1167,8 @@ def forward( torch.Size([1, 1, 4, 1025, 7680]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # num_concurrent_media and num_chunks are both currently 1 @@ -1354,10 +1346,8 @@ def forward( "USER: \nWhat's the content of the image? ASSISTANT: The image features a busy city street with a stop sign prominently displayed" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer diff --git a/src/transformers/models/llava/modeling_llava.py b/src/transformers/models/llava/modeling_llava.py index fdbaa3d29a93..ab58e176beb2 100644 --- a/src/transformers/models/llava/modeling_llava.py +++ b/src/transformers/models/llava/modeling_llava.py @@ -265,10 +265,8 @@ def forward( image_sizes: torch.Tensor = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, LlavaModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -452,10 +450,8 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "USER: \nWhat's the content of the image? ASSISTANT: The image features a busy city street with a stop sign prominently displayed" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/llava_next/modeling_llava_next.py b/src/transformers/models/llava_next/modeling_llava_next.py index 4ce50206fffd..a9fb4edda3df 100644 --- a/src/transformers/models/llava_next/modeling_llava_next.py +++ b/src/transformers/models/llava_next/modeling_llava_next.py @@ -465,10 +465,8 @@ def forward( Can be one of `"default"` or `"full"`. If `"default"`, the CLS token is removed from the vision features. If `"full"`, the full vision features are used. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -662,10 +660,8 @@ def forward( "[INST] \nWhat is shown in this image? [/INST] The image appears to be a radar chart, which is a type of multi-dimensional plot (...)" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer ) diff --git a/src/transformers/models/llava_next_video/modeling_llava_next_video.py b/src/transformers/models/llava_next_video/modeling_llava_next_video.py index 19d47b98495b..269d03f9f159 100644 --- a/src/transformers/models/llava_next_video/modeling_llava_next_video.py +++ b/src/transformers/models/llava_next_video/modeling_llava_next_video.py @@ -518,10 +518,8 @@ def forward( [`AutoImageProcessor`]. See [`LlavaNextVideoVideoProcessor.__call__`] for details. [`LlavaProcessor`] uses [`LlavaNextVideoVideoProcessor`] for processing videos. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict self.vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -831,10 +829,8 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "USER: \nWhat's the content of the image? ASSISTANT: The image shows a red stop sign on a pole, with a traditional Chinese archway (...)" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/llava_next_video/modular_llava_next_video.py b/src/transformers/models/llava_next_video/modular_llava_next_video.py index 490632888bc6..a4154f37508c 100644 --- a/src/transformers/models/llava_next_video/modular_llava_next_video.py +++ b/src/transformers/models/llava_next_video/modular_llava_next_video.py @@ -403,7 +403,7 @@ def forward( [`AutoImageProcessor`]. See [`LlavaNextVideoVideoProcessor.__call__`] for details. [`LlavaProcessor`] uses [`LlavaNextVideoVideoProcessor`] for processing videos. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -594,7 +594,7 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "USER: \nWhat's the content of the image? ASSISTANT: The image shows a red stop sign on a pole, with a traditional Chinese archway (...)" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/llava_onevision/modeling_llava_onevision.py b/src/transformers/models/llava_onevision/modeling_llava_onevision.py index 0e051c0b68d0..d7699e6dd389 100644 --- a/src/transformers/models/llava_onevision/modeling_llava_onevision.py +++ b/src/transformers/models/llava_onevision/modeling_llava_onevision.py @@ -547,10 +547,8 @@ def forward( Number of images in each sample. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -858,10 +856,8 @@ def forward( >>> processor.batch_decode(output, skip_special_tokens=True)[0] "user\n\nWhat is shown in this image?\nassistant\ncat" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/llava_onevision/modular_llava_onevision.py b/src/transformers/models/llava_onevision/modular_llava_onevision.py index ed476b6d93c4..a38b14c4a8b8 100644 --- a/src/transformers/models/llava_onevision/modular_llava_onevision.py +++ b/src/transformers/models/llava_onevision/modular_llava_onevision.py @@ -510,7 +510,7 @@ def forward( Number of images in each sample. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -684,7 +684,7 @@ def forward( >>> processor.batch_decode(output, skip_special_tokens=True)[0] "user\n\nWhat is shown in this image?\nassistant\ncat" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/longformer/modeling_longformer.py b/src/transformers/models/longformer/modeling_longformer.py index f1fec421e8ba..82417bee3c44 100755 --- a/src/transformers/models/longformer/modeling_longformer.py +++ b/src/transformers/models/longformer/modeling_longformer.py @@ -1590,10 +1590,8 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/longt5/modeling_flax_longt5.py b/src/transformers/models/longt5/modeling_flax_longt5.py index dee4afeadf72..724a2619bcd6 100644 --- a/src/transformers/models/longt5/modeling_flax_longt5.py +++ b/src/transformers/models/longt5/modeling_flax_longt5.py @@ -1735,9 +1735,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if decoder_input_ids is None: @@ -1835,9 +1833,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1903,9 +1899,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -2291,9 +2285,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/longt5/modeling_longt5.py b/src/transformers/models/longt5/modeling_longt5.py index 2ddcc4f3d4d1..08068309d893 100644 --- a/src/transformers/models/longt5/modeling_longt5.py +++ b/src/transformers/models/longt5/modeling_longt5.py @@ -1394,10 +1394,8 @@ def forward( cache_position=None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/luke/modeling_luke.py b/src/transformers/models/luke/modeling_luke.py index c01bb7453f9a..60afc65d5f3c 100644 --- a/src/transformers/models/luke/modeling_luke.py +++ b/src/transformers/models/luke/modeling_luke.py @@ -1009,10 +1009,8 @@ def forward( >>> word_last_hidden_state = outputs.last_hidden_state >>> entity_last_hidden_state = outputs.entity_last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/lxmert/modeling_lxmert.py b/src/transformers/models/lxmert/modeling_lxmert.py index 252fe2f40d07..fa4fd2634f6b 100644 --- a/src/transformers/models/lxmert/modeling_lxmert.py +++ b/src/transformers/models/lxmert/modeling_lxmert.py @@ -830,10 +830,8 @@ def forward( [What are attention masks?](../glossary#attention-mask) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/m2m_100/modeling_m2m_100.py b/src/transformers/models/m2m_100/modeling_m2m_100.py index 25ed975e4d52..653c4719fe89 100755 --- a/src/transformers/models/m2m_100/modeling_m2m_100.py +++ b/src/transformers/models/m2m_100/modeling_m2m_100.py @@ -828,10 +828,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -1039,10 +1037,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1287,10 +1283,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/mamba/modeling_mamba.py b/src/transformers/models/mamba/modeling_mamba.py index 0f6dfab81124..c9a8b91851a1 100644 --- a/src/transformers/models/mamba/modeling_mamba.py +++ b/src/transformers/models/mamba/modeling_mamba.py @@ -525,9 +525,7 @@ def forward( use_cache (`bool`, *optional*): If set to `True`, the `cache_params` is returned and can be used to quickly generate the next logits. """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else (self.config.use_cache if not self.training else False) return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/mamba2/modeling_mamba2.py b/src/transformers/models/mamba2/modeling_mamba2.py index 638ae9ccbb4a..f75ab4bbfc82 100644 --- a/src/transformers/models/mamba2/modeling_mamba2.py +++ b/src/transformers/models/mamba2/modeling_mamba2.py @@ -865,9 +865,7 @@ def forward( The position of the current input in the cache. This is used to ensure that the cache is correctly updated. If `cache_params` is passed, `cache_position` should also be passed. """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else (self.config.use_cache if not self.training else False) return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/marian/modeling_flax_marian.py b/src/transformers/models/marian/modeling_flax_marian.py index e1d9bea4fcdb..76aaab32e4c1 100644 --- a/src/transformers/models/marian/modeling_flax_marian.py +++ b/src/transformers/models/marian/modeling_flax_marian.py @@ -990,9 +990,7 @@ def encode( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1063,9 +1061,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1154,9 +1150,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1330,9 +1324,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/marian/modeling_marian.py b/src/transformers/models/marian/modeling_marian.py index 48390d047539..8d134d7f281e 100755 --- a/src/transformers/models/marian/modeling_marian.py +++ b/src/transformers/models/marian/modeling_marian.py @@ -778,10 +778,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -981,10 +979,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1294,10 +1290,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 26, 512] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1716,10 +1710,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/marian/modeling_tf_marian.py b/src/transformers/models/marian/modeling_tf_marian.py index 9b03424444f2..e0aac8c3e5a3 100644 --- a/src/transformers/models/marian/modeling_tf_marian.py +++ b/src/transformers/models/marian/modeling_tf_marian.py @@ -1168,9 +1168,7 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/markuplm/modeling_markuplm.py b/src/transformers/models/markuplm/modeling_markuplm.py index 8ce6b5ed5ec9..9f3b92f4eed9 100755 --- a/src/transformers/models/markuplm/modeling_markuplm.py +++ b/src/transformers/models/markuplm/modeling_markuplm.py @@ -800,10 +800,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 4, 768] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/mask2former/modeling_mask2former.py b/src/transformers/models/mask2former/modeling_mask2former.py index db942a3ac6ee..38ac873e9ff4 100644 --- a/src/transformers/models/mask2former/modeling_mask2former.py +++ b/src/transformers/models/mask2former/modeling_mask2former.py @@ -1146,10 +1146,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1276,10 +1274,8 @@ def forward( output_hidden_states=None, return_dict=None, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # Apply 1x1 convolution to reduce the channel dimension to d_model (256 by default) input_embeds = [] @@ -1822,10 +1818,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -2171,10 +2165,8 @@ def forward( output_attentions: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Mask2FormerModelOutput: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, _, height, width = pixel_values.shape @@ -2404,10 +2396,8 @@ def forward( torch.Size([338, 676]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/maskformer/modeling_maskformer.py b/src/transformers/models/maskformer/modeling_maskformer.py index 02f9848a8fba..e57cc1e381c7 100644 --- a/src/transformers/models/maskformer/modeling_maskformer.py +++ b/src/transformers/models/maskformer/modeling_maskformer.py @@ -708,10 +708,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1536,10 +1534,8 @@ def forward( if pixel_values is None: raise ValueError("You have to specify pixel_values") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, _, height, width = pixel_values.shape @@ -1753,10 +1749,8 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict raw_outputs = self.model( diff --git a/src/transformers/models/maskformer/modeling_maskformer_swin.py b/src/transformers/models/maskformer/modeling_maskformer_swin.py index f4845424850c..bf235a46203f 100644 --- a/src/transformers/models/maskformer/modeling_maskformer_swin.py +++ b/src/transformers/models/maskformer/modeling_maskformer_swin.py @@ -829,10 +829,8 @@ def forward( interpolate_pos_encoding=False, return_dict=None, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -915,10 +913,8 @@ def forward( return_dict: Optional[bool] = None, ) -> BackboneOutput: return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + outputs = self.model( pixel_values, output_hidden_states=True, output_attentions=output_attentions, return_dict=True diff --git a/src/transformers/models/mbart/modeling_flax_mbart.py b/src/transformers/models/mbart/modeling_flax_mbart.py index ad269860a959..dc0ac48b4efe 100644 --- a/src/transformers/models/mbart/modeling_flax_mbart.py +++ b/src/transformers/models/mbart/modeling_flax_mbart.py @@ -1064,9 +1064,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1136,9 +1134,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1227,9 +1223,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1401,9 +1395,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/mbart/modeling_mbart.py b/src/transformers/models/mbart/modeling_mbart.py index 18ad34026f41..4cce3b9deb3a 100755 --- a/src/transformers/models/mbart/modeling_mbart.py +++ b/src/transformers/models/mbart/modeling_mbart.py @@ -811,10 +811,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -1027,10 +1025,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1278,10 +1274,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1924,10 +1918,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/mbart/modeling_tf_mbart.py b/src/transformers/models/mbart/modeling_tf_mbart.py index 44ac6562a861..3f64655cb0fa 100644 --- a/src/transformers/models/mbart/modeling_tf_mbart.py +++ b/src/transformers/models/mbart/modeling_tf_mbart.py @@ -1183,9 +1183,7 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if decoder_input_ids is None and input_ids is not None: decoder_input_ids = shift_tokens_right(input_ids, self.config.pad_token_id) diff --git a/src/transformers/models/megatron_bert/modeling_megatron_bert.py b/src/transformers/models/megatron_bert/modeling_megatron_bert.py index abbda4085884..e8cdc97f6f66 100755 --- a/src/transformers/models/megatron_bert/modeling_megatron_bert.py +++ b/src/transformers/models/megatron_bert/modeling_megatron_bert.py @@ -806,10 +806,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/mgp_str/modeling_mgp_str.py b/src/transformers/models/mgp_str/modeling_mgp_str.py index 29e4d7981f8d..9d84f346d032 100644 --- a/src/transformers/models/mgp_str/modeling_mgp_str.py +++ b/src/transformers/models/mgp_str/modeling_mgp_str.py @@ -332,10 +332,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.FloatTensor], BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -422,10 +420,8 @@ def forward( >>> out_strs["generated_text"] '["ticket"]' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict mgp_outputs = self.mgp_str( diff --git a/src/transformers/models/mimi/modeling_mimi.py b/src/transformers/models/mimi/modeling_mimi.py index 9023c93433ab..107c8d5a7fe0 100644 --- a/src/transformers/models/mimi/modeling_mimi.py +++ b/src/transformers/models/mimi/modeling_mimi.py @@ -965,10 +965,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/minimax/modeling_minimax.py b/src/transformers/models/minimax/modeling_minimax.py index 1bf968e0361c..6e5d319e6da2 100644 --- a/src/transformers/models/minimax/modeling_minimax.py +++ b/src/transformers/models/minimax/modeling_minimax.py @@ -685,13 +685,11 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -949,14 +947,12 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/minimax/modular_minimax.py b/src/transformers/models/minimax/modular_minimax.py index 9b6fc12ae3de..c0b35adb87b6 100644 --- a/src/transformers/models/minimax/modular_minimax.py +++ b/src/transformers/models/minimax/modular_minimax.py @@ -492,7 +492,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) diff --git a/src/transformers/models/mistral/modeling_flax_mistral.py b/src/transformers/models/mistral/modeling_flax_mistral.py index 5b34a5144168..e36d1536fcde 100644 --- a/src/transformers/models/mistral/modeling_flax_mistral.py +++ b/src/transformers/models/mistral/modeling_flax_mistral.py @@ -470,9 +470,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/mistral/modeling_mistral.py b/src/transformers/models/mistral/modeling_mistral.py index 58e26c2943e7..c47b23b9d1c5 100644 --- a/src/transformers/models/mistral/modeling_mistral.py +++ b/src/transformers/models/mistral/modeling_mistral.py @@ -352,10 +352,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/mistral/modular_mistral.py b/src/transformers/models/mistral/modular_mistral.py index 7aa6b0b60efe..6758ccf7ac85 100644 --- a/src/transformers/models/mistral/modular_mistral.py +++ b/src/transformers/models/mistral/modular_mistral.py @@ -120,7 +120,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/mistral3/modeling_mistral3.py b/src/transformers/models/mistral3/modeling_mistral3.py index a5d12a115ba8..0d17e4812afe 100644 --- a/src/transformers/models/mistral3/modeling_mistral3.py +++ b/src/transformers/models/mistral3/modeling_mistral3.py @@ -310,10 +310,8 @@ def forward( image_sizes: torch.Tensor = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, Mistral3ModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -486,10 +484,8 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is the image?The image depicts two cats lying on a pink blanket." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/mistral3/modular_mistral3.py b/src/transformers/models/mistral3/modular_mistral3.py index 09cb5808712e..b58d2150fc1a 100644 --- a/src/transformers/models/mistral3/modular_mistral3.py +++ b/src/transformers/models/mistral3/modular_mistral3.py @@ -191,7 +191,7 @@ def forward( image_sizes: torch.Tensor = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, Mistral3ModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -311,7 +311,7 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is the image?The image depicts two cats lying on a pink blanket." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/mlcd/modeling_mlcd.py b/src/transformers/models/mlcd/modeling_mlcd.py index b20e9f107e09..3aa28161a26d 100644 --- a/src/transformers/models/mlcd/modeling_mlcd.py +++ b/src/transformers/models/mlcd/modeling_mlcd.py @@ -403,11 +403,9 @@ def forward( Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -470,11 +468,9 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -602,11 +598,9 @@ def forward( >>> print(f"Number of attention layers: {len(outputs.attentions)}") >>> print(f"Attention shape: {outputs.attentions[0].shape}") ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + return self.vision_model( pixel_values=pixel_values, diff --git a/src/transformers/models/mlcd/modular_mlcd.py b/src/transformers/models/mlcd/modular_mlcd.py index 600324532a30..f92efe9eea6d 100644 --- a/src/transformers/models/mlcd/modular_mlcd.py +++ b/src/transformers/models/mlcd/modular_mlcd.py @@ -346,7 +346,7 @@ def forward( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -405,7 +405,7 @@ def forward( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -518,7 +518,7 @@ def forward( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + return self.vision_model( pixel_values=pixel_values, diff --git a/src/transformers/models/mllama/modeling_mllama.py b/src/transformers/models/mllama/modeling_mllama.py index 91c8ffac8231..bef7ced51cd0 100644 --- a/src/transformers/models/mllama/modeling_mllama.py +++ b/src/transformers/models/mllama/modeling_mllama.py @@ -374,10 +374,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -1126,10 +1124,8 @@ def forward( torch.Size([1, 1, 4, 1025, 7680]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_concurrent_media, num_tiles, num_channels, height, width = pixel_values.shape @@ -1340,10 +1336,8 @@ def forward( torch.Size([1, 13, 4096]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1560,10 +1554,8 @@ def forward( I love the idea of snowflakes gently falling, each one ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) @@ -1686,10 +1678,8 @@ def forward( Output of the vision model, used for cross-attention. This tensor contains the processed image features that the language model will attend to. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): @@ -1883,10 +1873,8 @@ def forward( [', it would be:.\\nA stop sign in Chinatown.\\n'] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/mobilebert/modeling_mobilebert.py b/src/transformers/models/mobilebert/modeling_mobilebert.py index 7c34fd3d0b79..f6de810c0f1d 100644 --- a/src/transformers/models/mobilebert/modeling_mobilebert.py +++ b/src/transformers/models/mobilebert/modeling_mobilebert.py @@ -759,10 +759,8 @@ def forward( output_attentions: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/mobilenet_v1/modeling_mobilenet_v1.py b/src/transformers/models/mobilenet_v1/modeling_mobilenet_v1.py index b77dd75ebf90..0d3899b2de68 100755 --- a/src/transformers/models/mobilenet_v1/modeling_mobilenet_v1.py +++ b/src/transformers/models/mobilenet_v1/modeling_mobilenet_v1.py @@ -313,9 +313,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/mobilenet_v2/modeling_mobilenet_v2.py b/src/transformers/models/mobilenet_v2/modeling_mobilenet_v2.py index cc3173896c27..c86b168d280c 100755 --- a/src/transformers/models/mobilenet_v2/modeling_mobilenet_v2.py +++ b/src/transformers/models/mobilenet_v2/modeling_mobilenet_v2.py @@ -516,9 +516,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -754,9 +752,7 @@ def forward( >>> # logits are of shape (batch_size, num_labels, height, width) >>> logits = outputs.logits ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None and self.config.num_labels == 1: diff --git a/src/transformers/models/mobilevit/modeling_mobilevit.py b/src/transformers/models/mobilevit/modeling_mobilevit.py index 149eb9261ef1..f1319eac9de4 100755 --- a/src/transformers/models/mobilevit/modeling_mobilevit.py +++ b/src/transformers/models/mobilevit/modeling_mobilevit.py @@ -691,9 +691,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -967,9 +965,7 @@ def forward( >>> # logits are of shape (batch_size, num_labels, height, width) >>> logits = outputs.logits ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None and self.config.num_labels == 1: diff --git a/src/transformers/models/mobilevit/modeling_tf_mobilevit.py b/src/transformers/models/mobilevit/modeling_tf_mobilevit.py index 4acf1ff013ac..f7fc9ba30123 100644 --- a/src/transformers/models/mobilevit/modeling_tf_mobilevit.py +++ b/src/transformers/models/mobilevit/modeling_tf_mobilevit.py @@ -834,9 +834,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPooling]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # When running on CPU, `keras.layers.Conv2D` doesn't support `NCHW` format. @@ -1318,9 +1316,7 @@ def call( >>> # logits are of shape (batch_size, num_labels, height, width) >>> logits = outputs.logits ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None and not self.config.num_labels > 1: diff --git a/src/transformers/models/mobilevitv2/modeling_mobilevitv2.py b/src/transformers/models/mobilevitv2/modeling_mobilevitv2.py index 868c595dbace..76bb615d5941 100644 --- a/src/transformers/models/mobilevitv2/modeling_mobilevitv2.py +++ b/src/transformers/models/mobilevitv2/modeling_mobilevitv2.py @@ -642,9 +642,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -923,9 +921,7 @@ def forward( >>> # logits are of shape (batch_size, num_labels, height, width) >>> logits = outputs.logits ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None and self.config.num_labels == 1: diff --git a/src/transformers/models/modernbert/modeling_modernbert.py b/src/transformers/models/modernbert/modeling_modernbert.py index d984c523d0c8..01e6dedcd5c9 100644 --- a/src/transformers/models/modernbert/modeling_modernbert.py +++ b/src/transformers/models/modernbert/modeling_modernbert.py @@ -810,10 +810,8 @@ def forward( seq_len (`int`, *optional*): Sequence length of the input sequences including padding tokens. Used to pad the output tensors. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/modernbert/modular_modernbert.py b/src/transformers/models/modernbert/modular_modernbert.py index 72354be49248..4ca809b68aa7 100644 --- a/src/transformers/models/modernbert/modular_modernbert.py +++ b/src/transformers/models/modernbert/modular_modernbert.py @@ -939,7 +939,7 @@ def forward( seq_len (`int`, *optional*): Sequence length of the input sequences including padding tokens. Used to pad the output tensors. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/moonshine/modeling_moonshine.py b/src/transformers/models/moonshine/modeling_moonshine.py index c8b58869545c..130b5f83d2a6 100644 --- a/src/transformers/models/moonshine/modeling_moonshine.py +++ b/src/transformers/models/moonshine/modeling_moonshine.py @@ -576,10 +576,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if input_values is None: raise ValueError("You must specify input_values.") @@ -701,10 +699,8 @@ def forward( - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1083,10 +1079,8 @@ def forward( [1, 2, 288] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if encoder_outputs is None: diff --git a/src/transformers/models/moonshine/modular_moonshine.py b/src/transformers/models/moonshine/modular_moonshine.py index f620e98b1e2f..94c5d219c104 100644 --- a/src/transformers/models/moonshine/modular_moonshine.py +++ b/src/transformers/models/moonshine/modular_moonshine.py @@ -602,7 +602,7 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -707,7 +707,7 @@ def forward( - 0 for tokens that are **masked**. [What are attention masks?](../glossary#attention-mask) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -897,7 +897,7 @@ def forward( [1, 2, 288] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/moshi/modeling_moshi.py b/src/transformers/models/moshi/modeling_moshi.py index c0ef9c001471..94998afefc26 100644 --- a/src/transformers/models/moshi/modeling_moshi.py +++ b/src/transformers/models/moshi/modeling_moshi.py @@ -968,10 +968,8 @@ def forward( cache_position (`torch.Tensor`): Indices depicting the position of the input sequence tokens in the sequence. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1282,10 +1280,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1624,10 +1620,8 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/mpnet/modeling_mpnet.py b/src/transformers/models/mpnet/modeling_mpnet.py index efa74f191f13..fd1ec0653a53 100644 --- a/src/transformers/models/mpnet/modeling_mpnet.py +++ b/src/transformers/models/mpnet/modeling_mpnet.py @@ -455,10 +455,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/mpt/modeling_mpt.py b/src/transformers/models/mpt/modeling_mpt.py index bb2e89457a75..4f21740dd2fd 100644 --- a/src/transformers/models/mpt/modeling_mpt.py +++ b/src/transformers/models/mpt/modeling_mpt.py @@ -330,10 +330,8 @@ def forward( [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/mra/modeling_mra.py b/src/transformers/models/mra/modeling_mra.py index 7501c4d83062..6fc4f48fe6bb 100644 --- a/src/transformers/models/mra/modeling_mra.py +++ b/src/transformers/models/mra/modeling_mra.py @@ -875,9 +875,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCrossAttentions]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/mt5/modeling_mt5.py b/src/transformers/models/mt5/modeling_mt5.py index d2c7f5430c98..deef53bc1e25 100644 --- a/src/transformers/models/mt5/modeling_mt5.py +++ b/src/transformers/models/mt5/modeling_mt5.py @@ -954,10 +954,8 @@ def forward( torch.cuda.set_device(self.first_device) self.embed_tokens = self.embed_tokens.to(self.first_device) use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/musicgen/modeling_musicgen.py b/src/transformers/models/musicgen/modeling_musicgen.py index 0e9781aa5145..ab69925d6dd9 100644 --- a/src/transformers/models/musicgen/modeling_musicgen.py +++ b/src/transformers/models/musicgen/modeling_musicgen.py @@ -540,10 +540,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -821,10 +819,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/musicgen_melody/modeling_musicgen_melody.py b/src/transformers/models/musicgen_melody/modeling_musicgen_melody.py index a57955a7a70b..63259a9a3618 100644 --- a/src/transformers/models/musicgen_melody/modeling_musicgen_melody.py +++ b/src/transformers/models/musicgen_melody/modeling_musicgen_melody.py @@ -510,10 +510,8 @@ def forward( [What are attention masks?](../glossary#attention-mask) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -756,10 +754,8 @@ def forward( [What are attention masks?](../glossary#attention-mask) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/mvp/modeling_mvp.py b/src/transformers/models/mvp/modeling_mvp.py index c1db3afe7d57..03340b72eebe 100644 --- a/src/transformers/models/mvp/modeling_mvp.py +++ b/src/transformers/models/mvp/modeling_mvp.py @@ -619,10 +619,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -849,10 +847,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1096,10 +1092,8 @@ def forward( input_ids, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1787,10 +1781,8 @@ def forward( [1, 8, 50267] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/nemotron/modeling_nemotron.py b/src/transformers/models/nemotron/modeling_nemotron.py index 8c436fa83863..2a576d42035b 100644 --- a/src/transformers/models/nemotron/modeling_nemotron.py +++ b/src/transformers/models/nemotron/modeling_nemotron.py @@ -640,10 +640,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -933,10 +931,8 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/nllb_moe/modeling_nllb_moe.py b/src/transformers/models/nllb_moe/modeling_nllb_moe.py index 1f6ef1462c01..88a558fbbf47 100644 --- a/src/transformers/models/nllb_moe/modeling_nllb_moe.py +++ b/src/transformers/models/nllb_moe/modeling_nllb_moe.py @@ -965,10 +965,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict # retrieve input_ids and inputs_embeds @@ -1210,10 +1208,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.return_dict @@ -1692,7 +1688,7 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) diff --git a/src/transformers/models/nystromformer/modeling_nystromformer.py b/src/transformers/models/nystromformer/modeling_nystromformer.py index 17a3319de3ac..65702329d832 100755 --- a/src/transformers/models/nystromformer/modeling_nystromformer.py +++ b/src/transformers/models/nystromformer/modeling_nystromformer.py @@ -504,10 +504,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/olmoe/modeling_olmoe.py b/src/transformers/models/olmoe/modeling_olmoe.py index dbb81cddc219..5f3f61c99032 100644 --- a/src/transformers/models/olmoe/modeling_olmoe.py +++ b/src/transformers/models/olmoe/modeling_olmoe.py @@ -762,13 +762,11 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, MoeModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1078,13 +1076,11 @@ def forward( 'Hey, are you conscious? Can you talk to me?\nI’m not sure if you’re conscious of this, but I’m' ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py b/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py index c56c3415426e..ff6898f6cf33 100644 --- a/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py +++ b/src/transformers/models/omdet_turbo/modeling_omdet_turbo.py @@ -747,10 +747,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeddings @@ -933,8 +931,8 @@ def forward( output_attentions=None, output_hidden_states=None, ): - output_attentions = output_attentions if output_attentions is not None else self.output_attentions - output_hidden_states = output_hidden_states if output_hidden_states is not None else self.output_hidden_states + + origin_embedding_len = decoder_embeddings.shape[1] @@ -1333,10 +1331,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_features, vision_shapes, vision_shapes_list, level_start_index = self._get_encoder_input( @@ -1590,10 +1586,8 @@ def forward( if labels is not None: raise NotImplementedError("Training is not implemented yet") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict loss = None diff --git a/src/transformers/models/oneformer/modeling_oneformer.py b/src/transformers/models/oneformer/modeling_oneformer.py index bd740204b1b3..873d3ca0d98c 100644 --- a/src/transformers/models/oneformer/modeling_oneformer.py +++ b/src/transformers/models/oneformer/modeling_oneformer.py @@ -1231,10 +1231,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1374,10 +1372,8 @@ def forward( output_hidden_states=None, return_dict=None, ): - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # Then, apply 1x1 convolution to reduce the channel dimension to d_model (256 by default) sources = [] @@ -2929,10 +2925,8 @@ def forward( if pixel_values is None: raise ValueError("You have to specify pixel_values") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, _, height, width = pixel_values.shape @@ -3164,10 +3158,8 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/openai/modeling_openai.py b/src/transformers/models/openai/modeling_openai.py index e57d9f8a8964..e1fd89d330dd 100644 --- a/src/transformers/models/openai/modeling_openai.py +++ b/src/transformers/models/openai/modeling_openai.py @@ -453,10 +453,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/opt/modeling_flax_opt.py b/src/transformers/models/opt/modeling_flax_opt.py index d2f77ecbee26..e22af5e1c4b2 100644 --- a/src/transformers/models/opt/modeling_flax_opt.py +++ b/src/transformers/models/opt/modeling_flax_opt.py @@ -594,9 +594,7 @@ def __call__( deterministic: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: diff --git a/src/transformers/models/opt/modeling_opt.py b/src/transformers/models/opt/modeling_opt.py index 0dd16645b5b4..308d08b088bd 100644 --- a/src/transformers/models/opt/modeling_opt.py +++ b/src/transformers/models/opt/modeling_opt.py @@ -584,10 +584,8 @@ def forward( this tensor is not affected by padding. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -759,10 +757,8 @@ def forward( cache_position: Optional[torch.Tensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -864,10 +860,8 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious. I'm just a little bit of a weirdo." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/opt/modeling_tf_opt.py b/src/transformers/models/opt/modeling_tf_opt.py index f403d1cc5cc4..930064298503 100644 --- a/src/transformers/models/opt/modeling_tf_opt.py +++ b/src/transformers/models/opt/modeling_tf_opt.py @@ -634,9 +634,7 @@ def call( behaviors between training and evaluation). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -790,9 +788,7 @@ def call( **kwargs, ) -> Union[TFBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -870,9 +866,7 @@ def call( **kwargs, ) -> Union[TFBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1033,9 +1027,7 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/owlv2/modeling_owlv2.py b/src/transformers/models/owlv2/modeling_owlv2.py index 5d8065fdae0a..70d8caa87f55 100644 --- a/src/transformers/models/owlv2/modeling_owlv2.py +++ b/src/transformers/models/owlv2/modeling_owlv2.py @@ -641,10 +641,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -710,10 +708,8 @@ def forward( [`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for details. [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict input_shape = input_ids.size() @@ -834,10 +830,8 @@ def forward( interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Cast the input to the expected `dtype` @@ -1022,10 +1016,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use OWLv2 model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1075,10 +1067,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use OWLv2 model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1527,10 +1517,8 @@ def image_guided_detection( Detected similar object with confidence 0.966 at location [31.44, 463.65, 654.66, 471.07] Detected similar object with confidence 0.924 at location [30.93, 468.07, 635.35, 475.39] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict # Compute feature maps for the input and query images @@ -1637,10 +1625,8 @@ def forward( Detected a photo of a cat with confidence 0.614 at location [341.67, 23.39, 642.32, 371.35] Detected a photo of a cat with confidence 0.665 at location [6.75, 51.96, 326.62, 473.13] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict # Embed images and text queries diff --git a/src/transformers/models/owlvit/modeling_owlvit.py b/src/transformers/models/owlvit/modeling_owlvit.py index ff3b8a5b0424..cda5b92841b2 100644 --- a/src/transformers/models/owlvit/modeling_owlvit.py +++ b/src/transformers/models/owlvit/modeling_owlvit.py @@ -627,10 +627,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -695,10 +693,8 @@ def forward( [`PreTrainedTokenizer.encode`] and [`PreTrainedTokenizer.__call__`] for details. [What are input IDs?](../glossary#input-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict input_shape = input_ids.size() @@ -817,10 +813,8 @@ def forward( interpolate_pos_encoding: Optional[bool] = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Cast the input to the expected `dtype` @@ -1003,10 +997,8 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use OWL-ViT model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1056,10 +1048,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use OWL-ViT model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( @@ -1469,10 +1459,8 @@ def image_guided_detection( Detected similar object with confidence 0.856 at location [10.94, 50.4, 315.8, 471.39] Detected similar object with confidence 1.0 at location [334.84, 25.33, 636.16, 374.71] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict # Compute feature maps for the input and query images @@ -1579,10 +1567,8 @@ def forward( Detected a photo of a cat with confidence 0.707 at location [324.97, 20.44, 640.58, 373.29] Detected a photo of a cat with confidence 0.717 at location [1.46, 55.26, 315.55, 472.17] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.return_dict # Embed images and text queries diff --git a/src/transformers/models/paligemma/modeling_paligemma.py b/src/transformers/models/paligemma/modeling_paligemma.py index 4fd59aca1f5b..97bcc3eb29ef 100644 --- a/src/transformers/models/paligemma/modeling_paligemma.py +++ b/src/transformers/models/paligemma/modeling_paligemma.py @@ -306,10 +306,8 @@ def forward( if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict is_training = token_type_ids is not None and labels is not None @@ -484,10 +482,8 @@ def forward( >>> processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Where is the cat standing?\nsnow" ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/patchtst/modeling_patchtst.py b/src/transformers/models/patchtst/modeling_patchtst.py index 9817d0802080..f1a7402a8b00 100755 --- a/src/transformers/models/patchtst/modeling_patchtst.py +++ b/src/transformers/models/patchtst/modeling_patchtst.py @@ -744,10 +744,8 @@ def forward( return: `BaseModelOutput` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # Input embedding patch_input = self.embedder(patch_input) @@ -1194,10 +1192,8 @@ def forward( ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if past_observed_mask is None: past_observed_mask = torch.ones_like(past_values) diff --git a/src/transformers/models/pegasus/modeling_flax_pegasus.py b/src/transformers/models/pegasus/modeling_flax_pegasus.py index ddf0ae492407..734b084d8086 100644 --- a/src/transformers/models/pegasus/modeling_flax_pegasus.py +++ b/src/transformers/models/pegasus/modeling_flax_pegasus.py @@ -1007,9 +1007,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1080,9 +1078,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1171,9 +1167,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1348,9 +1342,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/pegasus/modeling_pegasus.py b/src/transformers/models/pegasus/modeling_pegasus.py index b8e69f957aa6..0a80e3aae559 100755 --- a/src/transformers/models/pegasus/modeling_pegasus.py +++ b/src/transformers/models/pegasus/modeling_pegasus.py @@ -790,10 +790,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -1032,10 +1030,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1311,10 +1307,8 @@ def forward( [1, 4, 1024] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1688,10 +1682,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/pegasus/modeling_tf_pegasus.py b/src/transformers/models/pegasus/modeling_tf_pegasus.py index 48f9e0b6953f..ec11449b086e 100644 --- a/src/transformers/models/pegasus/modeling_tf_pegasus.py +++ b/src/transformers/models/pegasus/modeling_tf_pegasus.py @@ -1183,9 +1183,7 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/pegasus_x/modeling_pegasus_x.py b/src/transformers/models/pegasus_x/modeling_pegasus_x.py index 90fe66934711..33c2b9c0ffba 100755 --- a/src/transformers/models/pegasus_x/modeling_pegasus_x.py +++ b/src/transformers/models/pegasus_x/modeling_pegasus_x.py @@ -1082,10 +1082,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -1294,10 +1292,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1553,10 +1549,8 @@ def forward( [1, 4, 1024] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/perceiver/modeling_perceiver.py b/src/transformers/models/perceiver/modeling_perceiver.py index c700f0a7ae14..7f78e55f77d7 100755 --- a/src/transformers/models/perceiver/modeling_perceiver.py +++ b/src/transformers/models/perceiver/modeling_perceiver.py @@ -788,10 +788,8 @@ def forward( >>> labels = torch.tensor([1]) >>> loss = criterion(logits, labels) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.input_preprocessor is not None: diff --git a/src/transformers/models/persimmon/modeling_persimmon.py b/src/transformers/models/persimmon/modeling_persimmon.py index 411c0612d90f..6e1e6b1430c7 100644 --- a/src/transformers/models/persimmon/modeling_persimmon.py +++ b/src/transformers/models/persimmon/modeling_persimmon.py @@ -457,10 +457,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -768,10 +766,8 @@ def forward( 'human: Hey, what should I eat for dinner?\n\ncat: 🐱\n\nhuman: 😐\n\n' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/phi/modeling_phi.py b/src/transformers/models/phi/modeling_phi.py index 7b63ebbc6dad..ec36311b6377 100644 --- a/src/transformers/models/phi/modeling_phi.py +++ b/src/transformers/models/phi/modeling_phi.py @@ -359,10 +359,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/phi/modular_phi.py b/src/transformers/models/phi/modular_phi.py index 0aef770b1e40..72b320575996 100644 --- a/src/transformers/models/phi/modular_phi.py +++ b/src/transformers/models/phi/modular_phi.py @@ -208,7 +208,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/phi3/modeling_phi3.py b/src/transformers/models/phi3/modeling_phi3.py index 31568b1aad8f..d4fc4e508eea 100644 --- a/src/transformers/models/phi3/modeling_phi3.py +++ b/src/transformers/models/phi3/modeling_phi3.py @@ -411,10 +411,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/phi4_multimodal/modeling_phi4_multimodal.py b/src/transformers/models/phi4_multimodal/modeling_phi4_multimodal.py index a9a902598c1f..a4fe24f51717 100644 --- a/src/transformers/models/phi4_multimodal/modeling_phi4_multimodal.py +++ b/src/transformers/models/phi4_multimodal/modeling_phi4_multimodal.py @@ -236,10 +236,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -551,10 +549,8 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + batch_size = pixel_values.size(0) if patch_attention_mask is None: @@ -1714,10 +1710,8 @@ def forward( audio_attention_mask (`torch.Tensor, *optional*): Attention mask for the audio inputs. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1896,10 +1890,8 @@ def forward( 'This is an example script .\n Certainly! Below is a sample script that demonstrates a simple task, such as calculating the sum' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py b/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py index 965cbb99a4b2..63d032bf39e4 100644 --- a/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py +++ b/src/transformers/models/phi4_multimodal/modular_phi4_multimodal.py @@ -672,7 +672,7 @@ def forward( patch_attention_mask: Optional[torch.BoolTensor] = None, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1528,7 +1528,7 @@ def forward( audio_attention_mask (`torch.Tensor, *optional*): Attention mask for the audio inputs. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -1686,7 +1686,7 @@ def forward( 'This is an example script .\n Certainly! Below is a sample script that demonstrates a simple task, such as calculating the sum' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/phimoe/modeling_phimoe.py b/src/transformers/models/phimoe/modeling_phimoe.py index 2fe0bb916e04..cc7c75bd6cb9 100644 --- a/src/transformers/models/phimoe/modeling_phimoe.py +++ b/src/transformers/models/phimoe/modeling_phimoe.py @@ -948,13 +948,11 @@ def forward( output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1305,14 +1303,12 @@ def forward( logger.warning( f"If you are not using the generate method, you may encounter nonsensical outputs after the {self.config.original_max_position_embeddings}th token, as the KV cache needs to be recomputed." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/pix2struct/modeling_pix2struct.py b/src/transformers/models/pix2struct/modeling_pix2struct.py index 63f07293d0f1..bfc5c1a91839 100644 --- a/src/transformers/models/pix2struct/modeling_pix2struct.py +++ b/src/transformers/models/pix2struct/modeling_pix2struct.py @@ -550,10 +550,8 @@ def forward( [1, 2048, 768] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if flattened_patches is None: @@ -1143,10 +1141,8 @@ def forward( ``` """ use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/pixtral/modeling_pixtral.py b/src/transformers/models/pixtral/modeling_pixtral.py index 998124a8da66..d4b5860e8e5e 100644 --- a/src/transformers/models/pixtral/modeling_pixtral.py +++ b/src/transformers/models/pixtral/modeling_pixtral.py @@ -361,10 +361,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None diff --git a/src/transformers/models/plbart/modeling_plbart.py b/src/transformers/models/plbart/modeling_plbart.py index a192fe70e238..b475832d447e 100644 --- a/src/transformers/models/plbart/modeling_plbart.py +++ b/src/transformers/models/plbart/modeling_plbart.py @@ -627,10 +627,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # retrieve input_ids and inputs_embeds @@ -959,10 +957,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1231,10 +1227,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1750,10 +1744,8 @@ def forward( True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/plbart/modular_plbart.py b/src/transformers/models/plbart/modular_plbart.py index 5881ab8ae902..7769a64c737f 100644 --- a/src/transformers/models/plbart/modular_plbart.py +++ b/src/transformers/models/plbart/modular_plbart.py @@ -361,7 +361,7 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/poolformer/modeling_poolformer.py b/src/transformers/models/poolformer/modeling_poolformer.py index ac6a27b3dd69..710d33adb316 100755 --- a/src/transformers/models/poolformer/modeling_poolformer.py +++ b/src/transformers/models/poolformer/modeling_poolformer.py @@ -287,9 +287,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/pop2piano/modeling_pop2piano.py b/src/transformers/models/pop2piano/modeling_pop2piano.py index 778e802e9678..8d02a86cbe11 100644 --- a/src/transformers/models/pop2piano/modeling_pop2piano.py +++ b/src/transformers/models/pop2piano/modeling_pop2piano.py @@ -704,10 +704,8 @@ def forward( cache_position=None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/prompt_depth_anything/modeling_prompt_depth_anything.py b/src/transformers/models/prompt_depth_anything/modeling_prompt_depth_anything.py index 2a463be89ff6..31774ac2636f 100644 --- a/src/transformers/models/prompt_depth_anything/modeling_prompt_depth_anything.py +++ b/src/transformers/models/prompt_depth_anything/modeling_prompt_depth_anything.py @@ -450,10 +450,8 @@ def forward( raise NotImplementedError("Training is not implemented yet") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + outputs = self.backbone.forward_with_filtered_kwargs( pixel_values, output_hidden_states=output_hidden_states, output_attentions=output_attentions diff --git a/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py b/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py index e5437b85cf66..a1871bc5b59f 100644 --- a/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py +++ b/src/transformers/models/prompt_depth_anything/modular_prompt_depth_anything.py @@ -295,7 +295,7 @@ def forward( output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + outputs = self.backbone.forward_with_filtered_kwargs( pixel_values, output_hidden_states=output_hidden_states, output_attentions=output_attentions diff --git a/src/transformers/models/prophetnet/modeling_prophetnet.py b/src/transformers/models/prophetnet/modeling_prophetnet.py index 17a48cd91c28..20e758b6742e 100644 --- a/src/transformers/models/prophetnet/modeling_prophetnet.py +++ b/src/transformers/models/prophetnet/modeling_prophetnet.py @@ -1142,10 +1142,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and inputs_embeds is None: @@ -1290,10 +1288,8 @@ def forward( >>> last_hidden_states = outputs.last_hidden_state ```""" use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None and inputs_embeds is None: @@ -1655,10 +1651,8 @@ def forward( >>> last_hidden_states_ngram = outputs.last_hidden_state_ngram # predict hidden states ```""" use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if encoder_outputs is None: diff --git a/src/transformers/models/pvt/modeling_pvt.py b/src/transformers/models/pvt/modeling_pvt.py index 2b776e83b877..4385074783c1 100755 --- a/src/transformers/models/pvt/modeling_pvt.py +++ b/src/transformers/models/pvt/modeling_pvt.py @@ -500,10 +500,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( diff --git a/src/transformers/models/pvt_v2/modeling_pvt_v2.py b/src/transformers/models/pvt_v2/modeling_pvt_v2.py index 33d3ccb7e931..b9007a95c100 100644 --- a/src/transformers/models/pvt_v2/modeling_pvt_v2.py +++ b/src/transformers/models/pvt_v2/modeling_pvt_v2.py @@ -441,10 +441,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( @@ -600,9 +598,7 @@ def forward( [1, 256, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + outputs = self.encoder( pixel_values=pixel_values, diff --git a/src/transformers/models/qwen2/modeling_qwen2.py b/src/transformers/models/qwen2/modeling_qwen2.py index d81faaf7a612..b68a62ac1345 100644 --- a/src/transformers/models/qwen2/modeling_qwen2.py +++ b/src/transformers/models/qwen2/modeling_qwen2.py @@ -357,10 +357,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/qwen2/modular_qwen2.py b/src/transformers/models/qwen2/modular_qwen2.py index ee80e6a24582..9a8a5034bc27 100644 --- a/src/transformers/models/qwen2/modular_qwen2.py +++ b/src/transformers/models/qwen2/modular_qwen2.py @@ -123,7 +123,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py b/src/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py index d6f2a0e4b5fc..6b99c42d43fc 100644 --- a/src/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py +++ b/src/transformers/models/qwen2_5_omni/modeling_qwen2_5_omni.py @@ -1691,10 +1691,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2007,10 +2005,8 @@ def forward( >>> response = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is None: @@ -2256,10 +2252,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2472,10 +2466,8 @@ def forward( "Generate the caption in English: Glass is breaking." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if attention_mask is not None and position_ids is None: diff --git a/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py b/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py index 78ae61168d8f..ebb3194e6bfe 100644 --- a/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py +++ b/src/transformers/models/qwen2_5_omni/modular_qwen2_5_omni.py @@ -2428,7 +2428,7 @@ def forward( >>> response = processor.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -2738,7 +2738,7 @@ def forward( "Generate the caption in English: Glass is breaking." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py b/src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py index 0001236c343a..756d31cdc1d7 100644 --- a/src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py +++ b/src/transformers/models/qwen2_5_vl/modeling_qwen2_5_vl.py @@ -922,10 +922,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1319,10 +1317,8 @@ def forward( The time interval (in seconds) for each grid along the temporal dimension in the 3D position IDs. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is None: @@ -1591,10 +1587,8 @@ def forward( "The image shows a street scene with a red stop sign in the foreground. In the background, there is a large red gate with Chinese characters ..." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs = self.model( input_ids=input_ids, diff --git a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py index adb745929289..0b9114e81910 100644 --- a/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py +++ b/src/transformers/models/qwen2_5_vl/modular_qwen2_5_vl.py @@ -638,7 +638,7 @@ def forward( The time interval (in seconds) for each grid along the temporal dimension in the 3D position IDs. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -826,7 +826,7 @@ def forward( "The image shows a street scene with a red stop sign in the foreground. In the background, there is a large red gate with Chinese characters ..." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py b/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py index 954d74081235..bf7a680e9651 100644 --- a/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py +++ b/src/transformers/models/qwen2_audio/modeling_qwen2_audio.py @@ -395,10 +395,8 @@ def forward( f"Qwen2Audio expects the mel input features to be of length {expected_seq_length}, but found {input_features.shape[-1]}. Make sure to pad the input mel features to {expected_seq_length}." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Ignore copy @@ -792,10 +790,8 @@ def forward( "Generate the caption in English: Glass is breaking." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict target_device = self.audio_tower.device diff --git a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py index ff0fb2dd526e..dc1077d097b8 100644 --- a/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py +++ b/src/transformers/models/qwen2_moe/modeling_qwen2_moe.py @@ -796,13 +796,11 @@ def forward( output_router_logits: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> MoeModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -1147,13 +1145,11 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py index 9bafbdc116f7..95ebc87980c9 100644 --- a/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py +++ b/src/transformers/models/qwen2_vl/modeling_qwen2_vl.py @@ -888,10 +888,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1245,10 +1243,8 @@ def forward( The rope index difference between sequence length and multimodal rope. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is None: @@ -1468,10 +1464,8 @@ def forward( "The image shows a street scene with a red stop sign in the foreground. In the background, there is a large red gate with Chinese characters ..." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs = self.model( input_ids=input_ids, diff --git a/src/transformers/models/qwen3/modeling_qwen3.py b/src/transformers/models/qwen3/modeling_qwen3.py index 5caeeab64b0e..4f5b39dfa4a5 100644 --- a/src/transformers/models/qwen3/modeling_qwen3.py +++ b/src/transformers/models/qwen3/modeling_qwen3.py @@ -382,10 +382,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py index d5674ba7d953..463d1676872c 100644 --- a/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py +++ b/src/transformers/models/qwen3_moe/modeling_qwen3_moe.py @@ -722,14 +722,12 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: MoeModelOutputWithPast = self.model( diff --git a/src/transformers/models/qwen3_moe/modular_qwen3_moe.py b/src/transformers/models/qwen3_moe/modular_qwen3_moe.py index 19780363f229..039508eb34db 100644 --- a/src/transformers/models/qwen3_moe/modular_qwen3_moe.py +++ b/src/transformers/models/qwen3_moe/modular_qwen3_moe.py @@ -271,7 +271,7 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_router_logits = ( output_router_logits if output_router_logits is not None else self.config.output_router_logits ) diff --git a/src/transformers/models/rag/modeling_rag.py b/src/transformers/models/rag/modeling_rag.py index 15d9fc61e882..69882df36f28 100644 --- a/src/transformers/models/rag/modeling_rag.py +++ b/src/transformers/models/rag/modeling_rag.py @@ -501,11 +501,9 @@ def forward( ```""" n_docs = n_docs if n_docs is not None else self.config.n_docs use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_retrieved = output_retrieved if output_retrieved is not None else self.config.output_retrieved + + + # whether retriever has to be used has_to_retrieve = ( diff --git a/src/transformers/models/recurrent_gemma/modeling_recurrent_gemma.py b/src/transformers/models/recurrent_gemma/modeling_recurrent_gemma.py index 47e79f34870a..6c99667670c1 100644 --- a/src/transformers/models/recurrent_gemma/modeling_recurrent_gemma.py +++ b/src/transformers/models/recurrent_gemma/modeling_recurrent_gemma.py @@ -612,9 +612,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -770,9 +768,7 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "What is your favorite condiment?" ```""" - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict output_hidden_states = True outputs = self.model( diff --git a/src/transformers/models/reformer/modeling_reformer.py b/src/transformers/models/reformer/modeling_reformer.py index ac9b71a64c24..f57d0e2222d2 100755 --- a/src/transformers/models/reformer/modeling_reformer.py +++ b/src/transformers/models/reformer/modeling_reformer.py @@ -1949,10 +1949,8 @@ def forward( up sequential decoding. """ use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/regnet/modeling_flax_regnet.py b/src/transformers/models/regnet/modeling_flax_regnet.py index 2cc3707fa51a..69882dd43651 100644 --- a/src/transformers/models/regnet/modeling_flax_regnet.py +++ b/src/transformers/models/regnet/modeling_flax_regnet.py @@ -603,9 +603,7 @@ def __call__( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ): - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -649,9 +647,7 @@ def __call__( output_hidden_states: bool = False, return_dict: bool = True, ) -> FlaxBaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, deterministic=deterministic) diff --git a/src/transformers/models/regnet/modeling_regnet.py b/src/transformers/models/regnet/modeling_regnet.py index aeb333e8b014..969468cef428 100644 --- a/src/transformers/models/regnet/modeling_regnet.py +++ b/src/transformers/models/regnet/modeling_regnet.py @@ -297,9 +297,7 @@ def __init__(self, config): def forward( self, pixel_values: Tensor, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None ) -> BaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values) diff --git a/src/transformers/models/regnet/modeling_tf_regnet.py b/src/transformers/models/regnet/modeling_tf_regnet.py index b06e64b3b406..81d153829813 100644 --- a/src/transformers/models/regnet/modeling_tf_regnet.py +++ b/src/transformers/models/regnet/modeling_tf_regnet.py @@ -395,9 +395,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> TFBaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, training=training) @@ -506,9 +504,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPoolingAndNoAttention, tuple[tf.Tensor]]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.regnet( @@ -574,9 +570,7 @@ def call( Labels for computing the image classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.regnet( diff --git a/src/transformers/models/rembert/modeling_rembert.py b/src/transformers/models/rembert/modeling_rembert.py index f6140b7a062e..4405fdc4b375 100755 --- a/src/transformers/models/rembert/modeling_rembert.py +++ b/src/transformers/models/rembert/modeling_rembert.py @@ -710,10 +710,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/resnet/modeling_flax_resnet.py b/src/transformers/models/resnet/modeling_flax_resnet.py index a2a9418b7cf2..ac4a3e94f5d9 100644 --- a/src/transformers/models/resnet/modeling_flax_resnet.py +++ b/src/transformers/models/resnet/modeling_flax_resnet.py @@ -494,9 +494,7 @@ def __call__( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ): - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -539,9 +537,7 @@ def __call__( output_hidden_states: bool = False, return_dict: bool = True, ) -> FlaxBaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, deterministic=deterministic) diff --git a/src/transformers/models/resnet/modeling_resnet.py b/src/transformers/models/resnet/modeling_resnet.py index c913bb1bdd99..996c2b11959f 100644 --- a/src/transformers/models/resnet/modeling_resnet.py +++ b/src/transformers/models/resnet/modeling_resnet.py @@ -281,9 +281,7 @@ def __init__(self, config): def forward( self, pixel_values: Tensor, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None ) -> BaseModelOutputWithPoolingAndNoAttention: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values) @@ -422,9 +420,7 @@ def forward( [1, 2048, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embedder(pixel_values) diff --git a/src/transformers/models/resnet/modeling_tf_resnet.py b/src/transformers/models/resnet/modeling_tf_resnet.py index f7c415f97b05..2ef260710481 100644 --- a/src/transformers/models/resnet/modeling_tf_resnet.py +++ b/src/transformers/models/resnet/modeling_tf_resnet.py @@ -420,9 +420,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 image layers can't use NCHW format when running on CPU. @@ -495,9 +493,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPoolingAndNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict resnet_outputs = self.resnet( diff --git a/src/transformers/models/roberta/modeling_flax_roberta.py b/src/transformers/models/roberta/modeling_flax_roberta.py index 3b46c0fa682f..fd4383c68de5 100644 --- a/src/transformers/models/roberta/modeling_flax_roberta.py +++ b/src/transformers/models/roberta/modeling_flax_roberta.py @@ -833,9 +833,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/roberta/modeling_roberta.py b/src/transformers/models/roberta/modeling_roberta.py index 39ce9d2e38b8..a665d917f58a 100644 --- a/src/transformers/models/roberta/modeling_roberta.py +++ b/src/transformers/models/roberta/modeling_roberta.py @@ -766,10 +766,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py b/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py index f65dc07bb165..add14852e368 100644 --- a/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py +++ b/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py @@ -837,9 +837,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py b/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py index 2ca11d1b15fe..f6c36c0a6ca1 100644 --- a/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py +++ b/src/transformers/models/roberta_prelayernorm/modeling_roberta_prelayernorm.py @@ -662,10 +662,8 @@ def forward( [What are token type IDs?](../glossary#token-type-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/roc_bert/modeling_roc_bert.py b/src/transformers/models/roc_bert/modeling_roc_bert.py index d1e199d5d4b2..35e185fdea61 100644 --- a/src/transformers/models/roc_bert/modeling_roc_bert.py +++ b/src/transformers/models/roc_bert/modeling_roc_bert.py @@ -857,10 +857,8 @@ def forward( [What are input IDs?](../glossary#input_pronunciation_ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/roformer/modeling_flax_roformer.py b/src/transformers/models/roformer/modeling_flax_roformer.py index de78eb4787c0..57327c97ba1f 100644 --- a/src/transformers/models/roformer/modeling_flax_roformer.py +++ b/src/transformers/models/roformer/modeling_flax_roformer.py @@ -656,9 +656,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/roformer/modeling_roformer.py b/src/transformers/models/roformer/modeling_roformer.py index 73afcf6858f7..e189a58d46d6 100644 --- a/src/transformers/models/roformer/modeling_roformer.py +++ b/src/transformers/models/roformer/modeling_roformer.py @@ -849,10 +849,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[BaseModelOutputWithPastAndCrossAttentions, tuple[torch.Tensor]]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/rt_detr/modeling_rt_detr.py b/src/transformers/models/rt_detr/modeling_rt_detr.py index 41669d9cf800..452b223a99c2 100644 --- a/src/transformers/models/rt_detr/modeling_rt_detr.py +++ b/src/transformers/models/rt_detr/modeling_rt_detr.py @@ -1241,10 +1241,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1382,10 +1380,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1671,10 +1667,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape @@ -1965,10 +1959,8 @@ def forward( Detected remote with confidence 0.951 at location [40.11, 73.44, 175.96, 118.48] Detected remote with confidence 0.924 at location [333.73, 76.58, 369.97, 186.99] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/rt_detr/modeling_rt_detr_resnet.py b/src/transformers/models/rt_detr/modeling_rt_detr_resnet.py index 17833c7037b8..98b86c21b658 100644 --- a/src/transformers/models/rt_detr/modeling_rt_detr_resnet.py +++ b/src/transformers/models/rt_detr/modeling_rt_detr_resnet.py @@ -358,9 +358,7 @@ def forward( [1, 2048, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + embedding_output = self.embedder(pixel_values) diff --git a/src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py b/src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py index b7ba9a11da5e..4affe3a4d0ec 100644 --- a/src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py +++ b/src/transformers/models/rt_detr_v2/modeling_rt_detr_v2.py @@ -1043,10 +1043,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1384,10 +1382,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1649,10 +1645,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape @@ -1958,10 +1952,8 @@ def forward( Detected remote with confidence 0.951 at location [40.11, 73.44, 175.96, 118.48] Detected remote with confidence 0.924 at location [333.73, 76.58, 369.97, 186.99] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/rwkv/modeling_rwkv.py b/src/transformers/models/rwkv/modeling_rwkv.py index edcea72ab8ab..38126757ddc7 100644 --- a/src/transformers/models/rwkv/modeling_rwkv.py +++ b/src/transformers/models/rwkv/modeling_rwkv.py @@ -560,10 +560,8 @@ def forward( use_cache (`bool`, *optional*): If set to `True`, the last state is returned and can be used to quickly generate the next logits. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else (self.config.use_cache if not self.training else False) return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/sam/modeling_sam.py b/src/transformers/models/sam/modeling_sam.py index c9d690d8e03c..21f65db8253d 100644 --- a/src/transformers/models/sam/modeling_sam.py +++ b/src/transformers/models/sam/modeling_sam.py @@ -407,10 +407,8 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + all_attentions = () @@ -1124,10 +1122,8 @@ def forward( self, pixel_values: Optional[torch.FloatTensor] = None, ) -> SamVisionEncoderOutput: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -1420,10 +1416,8 @@ def forward( ... ) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values is None and image_embeddings is None: raise ValueError("Either pixel_values or image_embeddings must be provided.") diff --git a/src/transformers/models/sam/modeling_tf_sam.py b/src/transformers/models/sam/modeling_tf_sam.py index c44dfdf80ac8..d9f071501257 100644 --- a/src/transformers/models/sam/modeling_tf_sam.py +++ b/src/transformers/models/sam/modeling_tf_sam.py @@ -436,9 +436,7 @@ def call( return_dict: Optional[bool] = None, ) -> Union[tuple, TFBaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_attentions = () @@ -1274,9 +1272,7 @@ def call( training: Optional[bool] = False, ) -> Union[tuple, TFSamVisionEncoderOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1583,9 +1579,7 @@ def call( **kwargs, ) -> TFSamImageSegmentationOutput | tuple[tf.Tensor]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None and image_embeddings is None: diff --git a/src/transformers/models/sam_hq/modeling_sam_hq.py b/src/transformers/models/sam_hq/modeling_sam_hq.py index 14f30fb4b401..68c542de4bd2 100644 --- a/src/transformers/models/sam_hq/modeling_sam_hq.py +++ b/src/transformers/models/sam_hq/modeling_sam_hq.py @@ -522,10 +522,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, SamHQVisionEncoderOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -833,10 +831,8 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + all_attentions = () @@ -1557,10 +1553,8 @@ def forward( ... ) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None and image_embeddings is None: diff --git a/src/transformers/models/sam_hq/modular_sam_hq.py b/src/transformers/models/sam_hq/modular_sam_hq.py index 62b927059d6f..e6ba9982597d 100644 --- a/src/transformers/models/sam_hq/modular_sam_hq.py +++ b/src/transformers/models/sam_hq/modular_sam_hq.py @@ -133,7 +133,7 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, SamHQVisionEncoderOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -582,7 +582,7 @@ def forward( ... ) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py b/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py index 4addfe42e0fc..b01cb401d29b 100755 --- a/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py +++ b/src/transformers/models/seamless_m4t/modeling_seamless_m4t.py @@ -1489,10 +1489,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Wav2Vec2BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_features is None: @@ -1633,10 +1631,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and self.is_t2u_encoder: @@ -1795,10 +1791,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1954,10 +1948,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2578,10 +2570,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2845,10 +2835,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3128,10 +3116,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3460,10 +3446,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3820,10 +3804,8 @@ def forward( config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py b/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py index 8891bd88777f..d5d964fafbd4 100644 --- a/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py +++ b/src/transformers/models/seamless_m4t_v2/modeling_seamless_m4t_v2.py @@ -1557,10 +1557,8 @@ def forward( return_dict: Optional[bool] = None, **kwargs, ) -> Union[tuple, Wav2Vec2BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_features is None: @@ -1702,10 +1700,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and self.is_t2u_encoder: @@ -1865,10 +1861,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2088,10 +2082,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # create padding mask for character lengths @@ -2204,10 +2196,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], Seq2SeqModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if encoder_outputs is None: @@ -2822,10 +2812,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3098,10 +3086,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3392,10 +3378,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -3765,10 +3749,8 @@ def forward( labels, self.config.pad_token_id, self.config.decoder_start_token_id ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -4165,10 +4147,8 @@ def forward( config.vocab_size]` (see `input_ids` docstring) Tokens with indices set to `-100` are ignored (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/segformer/modeling_segformer.py b/src/transformers/models/segformer/modeling_segformer.py index 5732a06b72a2..c8a908fb0608 100755 --- a/src/transformers/models/segformer/modeling_segformer.py +++ b/src/transformers/models/segformer/modeling_segformer.py @@ -475,10 +475,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( @@ -709,9 +707,7 @@ def forward( [1, 150, 128, 128] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and self.config.num_labels < 1: raise ValueError(f"Number of labels should be >=0: {self.config.num_labels}") diff --git a/src/transformers/models/segformer/modeling_tf_segformer.py b/src/transformers/models/segformer/modeling_tf_segformer.py index e3301b4dfc6a..5f759f8de217 100644 --- a/src/transformers/models/segformer/modeling_tf_segformer.py +++ b/src/transformers/models/segformer/modeling_tf_segformer.py @@ -586,9 +586,7 @@ def call( training: bool = False, ) -> Union[tuple, TFBaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # When running on CPU, `keras.layers.Conv2D` doesn't support `NCHW` format. @@ -984,9 +982,7 @@ def call( [1, 150, 128, 128] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + if labels is not None and not self.config.num_labels > 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/seggpt/modeling_seggpt.py b/src/transformers/models/seggpt/modeling_seggpt.py index 855096e12717..638a0228190b 100644 --- a/src/transformers/models/seggpt/modeling_seggpt.py +++ b/src/transformers/models/seggpt/modeling_seggpt.py @@ -729,10 +729,8 @@ def forward( [1, 56, 28, 1024] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict feature_ensemble = feature_ensemble if feature_ensemble is not None else False @@ -924,10 +922,8 @@ def forward( [170, 297] ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if bool_masked_pos is None: diff --git a/src/transformers/models/sew/modeling_sew.py b/src/transformers/models/sew/modeling_sew.py index 30949092bd0e..c0df614d7540 100644 --- a/src/transformers/models/sew/modeling_sew.py +++ b/src/transformers/models/sew/modeling_sew.py @@ -828,10 +828,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/sew/modular_sew.py b/src/transformers/models/sew/modular_sew.py index bb85bb1ea18a..cba5a721e773 100644 --- a/src/transformers/models/sew/modular_sew.py +++ b/src/transformers/models/sew/modular_sew.py @@ -417,7 +417,7 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/sew_d/modeling_sew_d.py b/src/transformers/models/sew_d/modeling_sew_d.py index 9bf50bb90e40..3731402a0c09 100644 --- a/src/transformers/models/sew_d/modeling_sew_d.py +++ b/src/transformers/models/sew_d/modeling_sew_d.py @@ -1351,10 +1351,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/siglip/modeling_siglip.py b/src/transformers/models/siglip/modeling_siglip.py index 6ca0c988860b..f5b6fdeb1ec4 100644 --- a/src/transformers/models/siglip/modeling_siglip.py +++ b/src/transformers/models/siglip/modeling_siglip.py @@ -613,10 +613,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -668,10 +666,8 @@ def forward( position_ids: Optional[torch.Tensor] = None, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if input_ids is None: raise ValueError("You have to specify input_ids") @@ -785,10 +781,8 @@ def forward( interpolate_pos_encoding: Optional[bool] = False, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + hidden_states = self.embeddings(pixel_values, interpolate_pos_encoding=interpolate_pos_encoding) @@ -956,10 +950,8 @@ def get_text_features( ... text_features = model.get_text_features(**inputs) ```""" # Use SigLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + text_outputs: BaseModelOutputWithPooling = self.text_model( input_ids=input_ids, @@ -1005,10 +997,8 @@ def get_image_features( ... image_features = model.get_image_features(**inputs) ```""" # Use SiglipModel's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1064,10 +1054,8 @@ def forward( 31.9% that image 0 is 'a photo of 2 cats' ```""" # Use SigLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1186,10 +1174,8 @@ def forward( >>> print("Predicted class:", model.config.id2label[predicted_class_idx]) Predicted class: LABEL_1 ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values, diff --git a/src/transformers/models/siglip2/modeling_siglip2.py b/src/transformers/models/siglip2/modeling_siglip2.py index 26b18b18f652..b37f8e77e337 100644 --- a/src/transformers/models/siglip2/modeling_siglip2.py +++ b/src/transformers/models/siglip2/modeling_siglip2.py @@ -441,10 +441,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -503,10 +501,8 @@ def forward( spatial_shapes (`torch.LongTensor` of shape `(batch_size, 2)`): Tensor containing the spatial dimensions (height, width) of the input images. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + hidden_states = self.embeddings(pixel_values, spatial_shapes) @@ -693,10 +689,8 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, ) -> BaseModelOutputWithPooling: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if input_ids is None: raise ValueError("You have to specify input_ids") @@ -1015,10 +1009,8 @@ def get_text_features( ... text_features = model.get_text_features(**inputs) ```""" # Use Siglip2 model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + text_outputs: BaseModelOutputWithPooling = self.text_model( input_ids=input_ids, @@ -1072,10 +1064,8 @@ def get_image_features( ``` """ # Use Siglip2Model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1139,10 +1129,8 @@ def forward( ``` """ # Use Siglip2 model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + vision_outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values=pixel_values, @@ -1269,10 +1257,8 @@ def forward( Predicted class: LABEL_1 ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + outputs: BaseModelOutputWithPooling = self.vision_model( pixel_values, diff --git a/src/transformers/models/siglip2/modular_siglip2.py b/src/transformers/models/siglip2/modular_siglip2.py index 4361fff5c285..d875f181d718 100644 --- a/src/transformers/models/siglip2/modular_siglip2.py +++ b/src/transformers/models/siglip2/modular_siglip2.py @@ -247,7 +247,7 @@ def forward( spatial_shapes (`torch.LongTensor` of shape `(batch_size, 2)`): Tensor containing the spatial dimensions (height, width) of the input images. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -395,7 +395,7 @@ def get_image_features( ``` """ # Use Siglip2Model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -460,7 +460,7 @@ def forward( ``` """ # Use Siglip2 model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -562,7 +562,7 @@ def forward( Predicted class: LABEL_1 ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/smolvlm/modeling_smolvlm.py b/src/transformers/models/smolvlm/modeling_smolvlm.py index 383450aae1f9..ba29d1464326 100644 --- a/src/transformers/models/smolvlm/modeling_smolvlm.py +++ b/src/transformers/models/smolvlm/modeling_smolvlm.py @@ -333,10 +333,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -410,10 +408,8 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size = pixel_values.size(0) @@ -713,10 +709,8 @@ def forward( image_hidden_states (`torch.FloatTensor` of shape `(batch_size, num_channels, image_size, image_size)`): The hidden states of the image encoder after modality projection. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -947,10 +941,8 @@ def forward( >>> print(generated_texts) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/smolvlm/modular_smolvlm.py b/src/transformers/models/smolvlm/modular_smolvlm.py index 6d9d19183825..4fde506aee83 100644 --- a/src/transformers/models/smolvlm/modular_smolvlm.py +++ b/src/transformers/models/smolvlm/modular_smolvlm.py @@ -269,7 +269,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, SmolVLMBaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py b/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py index d2d23d29007f..0a42cfc8218f 100644 --- a/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py +++ b/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py @@ -494,9 +494,7 @@ def encode( >>> encoder_outputs = model.encode(inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -574,9 +572,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -700,9 +696,7 @@ def __call__( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs diff --git a/src/transformers/models/speech_to_text/modeling_speech_to_text.py b/src/transformers/models/speech_to_text/modeling_speech_to_text.py index 8a44f497b56a..a1b41bdfe54e 100755 --- a/src/transformers/models/speech_to_text/modeling_speech_to_text.py +++ b/src/transformers/models/speech_to_text/modeling_speech_to_text.py @@ -646,10 +646,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict inputs_embeds = self.conv(input_features) inputs_embeds = self.embed_scale * inputs_embeds @@ -867,10 +865,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1167,10 +1163,8 @@ def forward( [1, 2, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py b/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py index fad6a026b152..c80a8ba807fe 100755 --- a/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py +++ b/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py @@ -1235,9 +1235,7 @@ def call( **kwargs, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/speecht5/modeling_speecht5.py b/src/transformers/models/speecht5/modeling_speecht5.py index fd4316775b7d..4e156c0844dd 100644 --- a/src/transformers/models/speecht5/modeling_speecht5.py +++ b/src/transformers/models/speecht5/modeling_speecht5.py @@ -1296,10 +1296,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # expand attention_mask @@ -1571,10 +1569,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -2042,10 +2038,8 @@ def forward( speaker_embeddings (`torch.FloatTensor` of shape `(batch_size, config.speaker_embedding_dim)`, *optional*): Tensor containing the speaker embeddings. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/splinter/modeling_splinter.py b/src/transformers/models/splinter/modeling_splinter.py index d1b09718e66b..e32e9420b71e 100755 --- a/src/transformers/models/splinter/modeling_splinter.py +++ b/src/transformers/models/splinter/modeling_splinter.py @@ -596,10 +596,8 @@ def forward( [What are position IDs?](../glossary#position-ids) """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/squeezebert/modeling_squeezebert.py b/src/transformers/models/squeezebert/modeling_squeezebert.py index ef72dfb1e819..d9e385aca66e 100644 --- a/src/transformers/models/squeezebert/modeling_squeezebert.py +++ b/src/transformers/models/squeezebert/modeling_squeezebert.py @@ -477,10 +477,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/stablelm/modeling_stablelm.py b/src/transformers/models/stablelm/modeling_stablelm.py index 14f503e41806..c92398b913d2 100755 --- a/src/transformers/models/stablelm/modeling_stablelm.py +++ b/src/transformers/models/stablelm/modeling_stablelm.py @@ -684,10 +684,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -996,10 +994,8 @@ def forward( 'human: Hey, what should I eat for dinner?\n\ncat: 🐱\n\nhuman: 😐\n\n' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( diff --git a/src/transformers/models/starcoder2/modeling_starcoder2.py b/src/transformers/models/starcoder2/modeling_starcoder2.py index 58f618b3d20f..d9896898de04 100644 --- a/src/transformers/models/starcoder2/modeling_starcoder2.py +++ b/src/transformers/models/starcoder2/modeling_starcoder2.py @@ -357,10 +357,8 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/src/transformers/models/starcoder2/modular_starcoder2.py b/src/transformers/models/starcoder2/modular_starcoder2.py index fa92b519b8c3..79417ebc76e6 100644 --- a/src/transformers/models/starcoder2/modular_starcoder2.py +++ b/src/transformers/models/starcoder2/modular_starcoder2.py @@ -176,7 +176,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/superglue/modeling_superglue.py b/src/transformers/models/superglue/modeling_superglue.py index 70b83805b799..9c1dd9e1ea18 100644 --- a/src/transformers/models/superglue/modeling_superglue.py +++ b/src/transformers/models/superglue/modeling_superglue.py @@ -777,10 +777,8 @@ def forward( if labels is not None: raise ValueError("SuperGlue is not trainable, no labels should be provided.") - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values.ndim != 5 or pixel_values.size(1) != 2: diff --git a/src/transformers/models/superpoint/modeling_superpoint.py b/src/transformers/models/superpoint/modeling_superpoint.py index bee932db5c8f..f05967d97e10 100644 --- a/src/transformers/models/superpoint/modeling_superpoint.py +++ b/src/transformers/models/superpoint/modeling_superpoint.py @@ -410,9 +410,7 @@ def forward( if labels is not None: raise ValueError("SuperPoint does not support training for now.") - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict pixel_values = self.extract_one_channel_pixel_values(pixel_values) diff --git a/src/transformers/models/swiftformer/modeling_swiftformer.py b/src/transformers/models/swiftformer/modeling_swiftformer.py index 10e626131a6e..853e2d9082d1 100644 --- a/src/transformers/models/swiftformer/modeling_swiftformer.py +++ b/src/transformers/models/swiftformer/modeling_swiftformer.py @@ -365,9 +365,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_hidden_states = (hidden_states,) if output_hidden_states else None @@ -424,9 +422,7 @@ def forward( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/swiftformer/modeling_tf_swiftformer.py b/src/transformers/models/swiftformer/modeling_tf_swiftformer.py index 612c2406a1d0..fd08d5df870f 100644 --- a/src/transformers/models/swiftformer/modeling_tf_swiftformer.py +++ b/src/transformers/models/swiftformer/modeling_tf_swiftformer.py @@ -550,9 +550,7 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple, TFBaseModelOutputWithNoAttention]: - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_hidden_states = (hidden_states,) if output_hidden_states else None @@ -659,9 +657,7 @@ def call( ) -> Union[tuple, TFBaseModelOutputWithNoAttention]: r""" """ - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 image layers can't use NCHW format when running on CPU. diff --git a/src/transformers/models/swin/modeling_swin.py b/src/transformers/models/swin/modeling_swin.py index 9713105de9c8..8577c9d968e6 100644 --- a/src/transformers/models/swin/modeling_swin.py +++ b/src/transformers/models/swin/modeling_swin.py @@ -976,10 +976,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1277,10 +1275,8 @@ def forward( [1, 768, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output, input_dimensions = self.embeddings(pixel_values) diff --git a/src/transformers/models/swin/modeling_tf_swin.py b/src/transformers/models/swin/modeling_tf_swin.py index 5b9f9f272ff4..276a1ff0d19d 100644 --- a/src/transformers/models/swin/modeling_tf_swin.py +++ b/src/transformers/models/swin/modeling_tf_swin.py @@ -1242,9 +1242,7 @@ def call( training: bool = False, ) -> Union[TFSwinModelOutput, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1342,9 +1340,7 @@ def call( Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/swin2sr/modeling_swin2sr.py b/src/transformers/models/swin2sr/modeling_swin2sr.py index 4385dc59230c..e2ae37f940ea 100644 --- a/src/transformers/models/swin2sr/modeling_swin2sr.py +++ b/src/transformers/models/swin2sr/modeling_swin2sr.py @@ -811,10 +811,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed diff --git a/src/transformers/models/swinv2/modeling_swinv2.py b/src/transformers/models/swinv2/modeling_swinv2.py index 9e6b401f7df8..f08f5cf07bd9 100644 --- a/src/transformers/models/swinv2/modeling_swinv2.py +++ b/src/transformers/models/swinv2/modeling_swinv2.py @@ -1033,10 +1033,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1330,10 +1328,8 @@ def forward( [1, 2048, 7, 7] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output, input_dimensions = self.embeddings(pixel_values) diff --git a/src/transformers/models/switch_transformers/modeling_switch_transformers.py b/src/transformers/models/switch_transformers/modeling_switch_transformers.py index cd0e690304b8..841edf37ac1f 100644 --- a/src/transformers/models/switch_transformers/modeling_switch_transformers.py +++ b/src/transformers/models/switch_transformers/modeling_switch_transformers.py @@ -913,10 +913,8 @@ def forward( cache_position=None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/t5/modeling_flax_t5.py b/src/transformers/models/t5/modeling_flax_t5.py index c829a084e4d9..734d77924f90 100644 --- a/src/transformers/models/t5/modeling_flax_t5.py +++ b/src/transformers/models/t5/modeling_flax_t5.py @@ -997,9 +997,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if decoder_input_ids is None: @@ -1097,9 +1095,7 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1165,9 +1161,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1466,9 +1460,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1643,9 +1635,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/t5/modeling_t5.py b/src/transformers/models/t5/modeling_t5.py index 057a52dd6995..38a0fa25f252 100644 --- a/src/transformers/models/t5/modeling_t5.py +++ b/src/transformers/models/t5/modeling_t5.py @@ -969,10 +969,8 @@ def forward( torch.cuda.set_device(self.first_device) self.embed_tokens = self.embed_tokens.to(self.first_device) use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/table_transformer/modeling_table_transformer.py b/src/transformers/models/table_transformer/modeling_table_transformer.py index 5d5d18ecb22c..9be105c84e54 100644 --- a/src/transformers/models/table_transformer/modeling_table_transformer.py +++ b/src/transformers/models/table_transformer/modeling_table_transformer.py @@ -816,10 +816,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -947,10 +945,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1134,10 +1130,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 15, 256] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/tapas/modeling_tapas.py b/src/transformers/models/tapas/modeling_tapas.py index 88e05f5f8fd3..80aba7c0d592 100644 --- a/src/transformers/models/tapas/modeling_tapas.py +++ b/src/transformers/models/tapas/modeling_tapas.py @@ -813,10 +813,8 @@ class for more info. >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/textnet/modeling_textnet.py b/src/transformers/models/textnet/modeling_textnet.py index 02c0fe4582c6..f897a2717f5f 100644 --- a/src/transformers/models/textnet/modeling_textnet.py +++ b/src/transformers/models/textnet/modeling_textnet.py @@ -247,9 +247,7 @@ def forward( self, pixel_values: Tensor, output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None ) -> Union[tuple[Any, list[Any]], tuple[Any], BaseModelOutputWithPoolingAndNoAttention]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + hidden_state = self.stem(pixel_values) @@ -402,9 +400,7 @@ def forward( >>> outputs = model(**inputs) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + outputs = self.textnet(pixel_values, output_hidden_states=True, return_dict=return_dict) diff --git a/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py b/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py index 0f89eb8a6f24..756ab9e0422c 100644 --- a/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py +++ b/src/transformers/models/time_series_transformer/modeling_time_series_transformer.py @@ -815,10 +815,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.value_embedding(inputs_embeds) @@ -990,10 +988,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1410,10 +1406,8 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/timesformer/modeling_timesformer.py b/src/transformers/models/timesformer/modeling_timesformer.py index 7610ab48e3fb..38f4f96af5a6 100644 --- a/src/transformers/models/timesformer/modeling_timesformer.py +++ b/src/transformers/models/timesformer/modeling_timesformer.py @@ -586,10 +586,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 1569, 768] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/timm_backbone/modeling_timm_backbone.py b/src/transformers/models/timm_backbone/modeling_timm_backbone.py index 2fab2df53a27..393962f1698e 100644 --- a/src/transformers/models/timm_backbone/modeling_timm_backbone.py +++ b/src/transformers/models/timm_backbone/modeling_timm_backbone.py @@ -127,10 +127,8 @@ def forward( **kwargs, ) -> Union[BackboneOutput, tuple[Tensor, ...]]: return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + if output_attentions: raise ValueError("Cannot output attentions for timm backbones at the moment") diff --git a/src/transformers/models/timm_wrapper/modeling_timm_wrapper.py b/src/transformers/models/timm_wrapper/modeling_timm_wrapper.py index 57c96aef27be..6431e554cb43 100644 --- a/src/transformers/models/timm_wrapper/modeling_timm_wrapper.py +++ b/src/transformers/models/timm_wrapper/modeling_timm_wrapper.py @@ -172,10 +172,8 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + do_pooling = do_pooling if do_pooling is not None else self.config.do_pooling if output_attentions: @@ -290,10 +288,8 @@ def forward( ``` """ return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + if output_attentions: raise ValueError("Cannot set `output_attentions` for timm models.") diff --git a/src/transformers/models/trocr/modeling_trocr.py b/src/transformers/models/trocr/modeling_trocr.py index 9aa1d28bce03..3529ea26ad18 100644 --- a/src/transformers/models/trocr/modeling_trocr.py +++ b/src/transformers/models/trocr/modeling_trocr.py @@ -561,10 +561,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -830,10 +828,8 @@ def forward( 'industry, " Mr. Brown commented icily. " Let us have a' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/tvp/modeling_tvp.py b/src/transformers/models/tvp/modeling_tvp.py index 302a9075c1ab..fca75c0a66c5 100644 --- a/src/transformers/models/tvp/modeling_tvp.py +++ b/src/transformers/models/tvp/modeling_tvp.py @@ -499,10 +499,8 @@ def forward( return_dict: Optional[bool] = None, ): return_dict = return_dict if return_dict is not None else self.config.return_dict - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + all_hidden_states = () all_attentions = () diff --git a/src/transformers/models/udop/modeling_udop.py b/src/transformers/models/udop/modeling_udop.py index 41efd964c99b..5ae0b8d28b70 100644 --- a/src/transformers/models/udop/modeling_udop.py +++ b/src/transformers/models/udop/modeling_udop.py @@ -1159,10 +1159,8 @@ def forward( cache_position=None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # input embeddings processing @@ -2032,10 +2030,8 @@ def forward( >>> outputs = model(**encoding) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_outputs = self.encoder( diff --git a/src/transformers/models/umt5/modeling_umt5.py b/src/transformers/models/umt5/modeling_umt5.py index bf0a0a80bbac..4ae0676cf242 100644 --- a/src/transformers/models/umt5/modeling_umt5.py +++ b/src/transformers/models/umt5/modeling_umt5.py @@ -652,10 +652,8 @@ def forward( cache_position=None, ): use_cache = use_cache if use_cache is not None else self.config.use_cache - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/unispeech/modeling_unispeech.py b/src/transformers/models/unispeech/modeling_unispeech.py index 02fa44891f51..8bb99a629752 100755 --- a/src/transformers/models/unispeech/modeling_unispeech.py +++ b/src/transformers/models/unispeech/modeling_unispeech.py @@ -1115,10 +1115,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/unispeech/modular_unispeech.py b/src/transformers/models/unispeech/modular_unispeech.py index fee5a3b6cc76..6644b091cb6d 100644 --- a/src/transformers/models/unispeech/modular_unispeech.py +++ b/src/transformers/models/unispeech/modular_unispeech.py @@ -257,7 +257,7 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/unispeech_sat/modeling_unispeech_sat.py b/src/transformers/models/unispeech_sat/modeling_unispeech_sat.py index 72375f8e904f..d24ed1e253b2 100755 --- a/src/transformers/models/unispeech_sat/modeling_unispeech_sat.py +++ b/src/transformers/models/unispeech_sat/modeling_unispeech_sat.py @@ -1117,10 +1117,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/unispeech_sat/modular_unispeech_sat.py b/src/transformers/models/unispeech_sat/modular_unispeech_sat.py index b4c209e58f77..13fb4f6e537c 100644 --- a/src/transformers/models/unispeech_sat/modular_unispeech_sat.py +++ b/src/transformers/models/unispeech_sat/modular_unispeech_sat.py @@ -266,7 +266,7 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/upernet/modeling_upernet.py b/src/transformers/models/upernet/modeling_upernet.py index 30dc93a86f42..373122ee75cc 100644 --- a/src/transformers/models/upernet/modeling_upernet.py +++ b/src/transformers/models/upernet/modeling_upernet.py @@ -340,10 +340,8 @@ def forward( raise ValueError("The number of labels should be greater than one") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + outputs = self.backbone.forward_with_filtered_kwargs( pixel_values, output_hidden_states=output_hidden_states, output_attentions=output_attentions diff --git a/src/transformers/models/video_llava/modeling_video_llava.py b/src/transformers/models/video_llava/modeling_video_llava.py index ce8f637d0b1c..e1ce46b3c62c 100644 --- a/src/transformers/models/video_llava/modeling_video_llava.py +++ b/src/transformers/models/video_llava/modeling_video_llava.py @@ -324,10 +324,8 @@ def forward( [`AutoImageProcessor`]. See [`VideoLlavaImageProcessor.__call__`] for details ([]`LlavaProcessor`] uses [`VideoLlavaImageProcessor`] for processing videos). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -578,10 +576,8 @@ def forward( ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer diff --git a/src/transformers/models/videomae/modeling_videomae.py b/src/transformers/models/videomae/modeling_videomae.py index 1c5f49ed7fe3..37215043e4ee 100755 --- a/src/transformers/models/videomae/modeling_videomae.py +++ b/src/transformers/models/videomae/modeling_videomae.py @@ -623,10 +623,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 1568, 768] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed diff --git a/src/transformers/models/vilt/modeling_vilt.py b/src/transformers/models/vilt/modeling_vilt.py index 974c9c86ae89..35f743839dd5 100755 --- a/src/transformers/models/vilt/modeling_vilt.py +++ b/src/transformers/models/vilt/modeling_vilt.py @@ -645,10 +645,8 @@ def forward( >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: @@ -1194,10 +1192,8 @@ def forward( >>> print("Predicted answer:", model.config.id2label[idx]) Predicted answer: True ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is not None and pixel_values.ndim == 4: diff --git a/src/transformers/models/vipllava/modeling_vipllava.py b/src/transformers/models/vipllava/modeling_vipllava.py index 55b90db70cb3..c4b59a375a2e 100644 --- a/src/transformers/models/vipllava/modeling_vipllava.py +++ b/src/transformers/models/vipllava/modeling_vipllava.py @@ -235,10 +235,8 @@ def forward( The vision feature layer, or the list of indexes of the layers to select the vision feature. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layers = ( vision_feature_layers if vision_feature_layers is not None else self.config.vision_feature_layers @@ -404,10 +402,8 @@ def forward( The image features a brown and white cat sitting on a green surface, with a red ball in its ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layers = ( vision_feature_layers if vision_feature_layers is not None else self.config.vision_feature_layers diff --git a/src/transformers/models/vipllava/modular_vipllava.py b/src/transformers/models/vipllava/modular_vipllava.py index 55dd06482b5a..e74f01974bb2 100644 --- a/src/transformers/models/vipllava/modular_vipllava.py +++ b/src/transformers/models/vipllava/modular_vipllava.py @@ -123,7 +123,7 @@ def forward( The vision feature layer, or the list of indexes of the layers to select the vision feature. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -239,7 +239,7 @@ def forward( The image features a brown and white cat sitting on a green surface, with a red ball in its ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/vision_encoder_decoder/modeling_flax_vision_encoder_decoder.py b/src/transformers/models/vision_encoder_decoder/modeling_flax_vision_encoder_decoder.py index c5467852c282..fcfc5440c17c 100644 --- a/src/transformers/models/vision_encoder_decoder/modeling_flax_vision_encoder_decoder.py +++ b/src/transformers/models/vision_encoder_decoder/modeling_flax_vision_encoder_decoder.py @@ -427,9 +427,7 @@ def encode( >>> encoder_outputs = model.encode(pixel_values) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # `FlaxViTModel` expects channel first format, but `FlaxViTModule` expects channel last format. @@ -512,9 +510,7 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -646,9 +642,7 @@ def __call__( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs diff --git a/src/transformers/models/vision_text_dual_encoder/modeling_flax_vision_text_dual_encoder.py b/src/transformers/models/vision_text_dual_encoder/modeling_flax_vision_text_dual_encoder.py index 44c5e6c1729b..d82df0dac444 100644 --- a/src/transformers/models/vision_text_dual_encoder/modeling_flax_vision_text_dual_encoder.py +++ b/src/transformers/models/vision_text_dual_encoder/modeling_flax_vision_text_dual_encoder.py @@ -281,9 +281,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/visual_bert/modeling_visual_bert.py b/src/transformers/models/visual_bert/modeling_visual_bert.py index 8d82926dc651..7c6f02a319e9 100755 --- a/src/transformers/models/visual_bert/modeling_visual_bert.py +++ b/src/transformers/models/visual_bert/modeling_visual_bert.py @@ -658,10 +658,8 @@ def forward( last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/vit/modeling_flax_vit.py b/src/transformers/models/vit/modeling_flax_vit.py index d62ef1b6b928..19fc31152eb0 100644 --- a/src/transformers/models/vit/modeling_flax_vit.py +++ b/src/transformers/models/vit/modeling_flax_vit.py @@ -482,9 +482,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/vit/modeling_vit.py b/src/transformers/models/vit/modeling_vit.py index 351dac20e3ad..b9f7ebc34aa0 100644 --- a/src/transformers/models/vit/modeling_vit.py +++ b/src/transformers/models/vit/modeling_vit.py @@ -527,10 +527,8 @@ def forward( bool_masked_pos (`torch.BoolTensor` of shape `(batch_size, num_patches)`, *optional*): Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/vit_mae/modeling_vit_mae.py b/src/transformers/models/vit_mae/modeling_vit_mae.py index f9611dc709ab..47c964631865 100755 --- a/src/transformers/models/vit_mae/modeling_vit_mae.py +++ b/src/transformers/models/vit_mae/modeling_vit_mae.py @@ -713,10 +713,8 @@ def forward( >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/vit_msn/modeling_vit_msn.py b/src/transformers/models/vit_msn/modeling_vit_msn.py index 0fbfd7e12d99..b0144907c516 100644 --- a/src/transformers/models/vit_msn/modeling_vit_msn.py +++ b/src/transformers/models/vit_msn/modeling_vit_msn.py @@ -536,10 +536,8 @@ def forward( ... outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/vitdet/modeling_vitdet.py b/src/transformers/models/vitdet/modeling_vitdet.py index b216d08c5149..f3ab3913626b 100644 --- a/src/transformers/models/vitdet/modeling_vitdet.py +++ b/src/transformers/models/vitdet/modeling_vitdet.py @@ -703,10 +703,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 768, 14, 14] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -788,10 +786,8 @@ def forward( [1, 768, 14, 14] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + embedding_output = self.embeddings(pixel_values) diff --git a/src/transformers/models/vitmatte/modeling_vitmatte.py b/src/transformers/models/vitmatte/modeling_vitmatte.py index a264695efbeb..c7572efb62a8 100644 --- a/src/transformers/models/vitmatte/modeling_vitmatte.py +++ b/src/transformers/models/vitmatte/modeling_vitmatte.py @@ -268,10 +268,8 @@ def forward( torch.Size([1, 1, 640, 960]) ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + loss = None if labels is not None: diff --git a/src/transformers/models/vitpose/modeling_vitpose.py b/src/transformers/models/vitpose/modeling_vitpose.py index 350c7c848739..f277825f1bef 100644 --- a/src/transformers/models/vitpose/modeling_vitpose.py +++ b/src/transformers/models/vitpose/modeling_vitpose.py @@ -255,10 +255,8 @@ def forward( ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + loss = None if labels is not None: diff --git a/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py b/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py index b6880ec4d003..afd9cdc33dd8 100644 --- a/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py +++ b/src/transformers/models/vitpose_backbone/modeling_vitpose_backbone.py @@ -483,10 +483,8 @@ def forward( >>> dataset_index = torch.tensor([1]) >>> outputs = model(pixel_values, dataset_index) ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # Prepare head mask if needed diff --git a/src/transformers/models/vits/modeling_vits.py b/src/transformers/models/vits/modeling_vits.py index 5f67a0bcb7a7..ddab00f70504 100644 --- a/src/transformers/models/vits/modeling_vits.py +++ b/src/transformers/models/vits/modeling_vits.py @@ -1335,10 +1335,8 @@ def forward( torch.Size([1, 45824]) ``` """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None: diff --git a/src/transformers/models/vivit/modeling_vivit.py b/src/transformers/models/vivit/modeling_vivit.py index ab93d5cf6e80..32ba35d7ff93 100755 --- a/src/transformers/models/vivit/modeling_vivit.py +++ b/src/transformers/models/vivit/modeling_vivit.py @@ -595,10 +595,8 @@ def forward( >>> list(last_hidden_states.shape) [1, 3137, 768] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/vjepa2/modeling_vjepa2.py b/src/transformers/models/vjepa2/modeling_vjepa2.py index bb139d4aeee4..005d6ee07f71 100644 --- a/src/transformers/models/vjepa2/modeling_vjepa2.py +++ b/src/transformers/models/vjepa2/modeling_vjepa2.py @@ -1105,10 +1105,8 @@ def forward( skip_predictor (bool): flag to skip the predictor forward, useful if you just need the encoder outputs """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if pixel_values_videos is None: raise ValueError("You have to specify pixel_values_videos") diff --git a/src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py b/src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py index ca5eb700ee8f..f75e08557f83 100644 --- a/src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py +++ b/src/transformers/models/wav2vec2/modeling_flax_wav2vec2.py @@ -900,9 +900,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_values.shape @@ -1337,9 +1335,7 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_values.shape diff --git a/src/transformers/models/wav2vec2/modeling_wav2vec2.py b/src/transformers/models/wav2vec2/modeling_wav2vec2.py index b7d764f712cf..7c33f61ff259 100755 --- a/src/transformers/models/wav2vec2/modeling_wav2vec2.py +++ b/src/transformers/models/wav2vec2/modeling_wav2vec2.py @@ -1505,10 +1505,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/wav2vec2_bert/configuration_wav2vec2_bert.py b/src/transformers/models/wav2vec2_bert/configuration_wav2vec2_bert.py index 2462976cfbbe..1b7fce69a6e9 100644 --- a/src/transformers/models/wav2vec2_bert/configuration_wav2vec2_bert.py +++ b/src/transformers/models/wav2vec2_bert/configuration_wav2vec2_bert.py @@ -288,7 +288,7 @@ def __init__( self.adapter_stride = adapter_stride self.num_adapter_layers = num_adapter_layers self.adapter_act = adapter_act - self.output_hidden_size = output_hidden_size if output_hidden_size is not None else hidden_size + self. if use_intermediate_ffn_before_adapter and not add_adapter: raise ValueError("`use_intermediate_ffn_before_adapter` is `True` but `add_adapter` is `False`.") self.use_intermediate_ffn_before_adapter = use_intermediate_ffn_before_adapter diff --git a/src/transformers/models/wav2vec2_bert/modeling_wav2vec2_bert.py b/src/transformers/models/wav2vec2_bert/modeling_wav2vec2_bert.py index b8a60f3d3d0e..8cb3ebcc6494 100644 --- a/src/transformers/models/wav2vec2_bert/modeling_wav2vec2_bert.py +++ b/src/transformers/models/wav2vec2_bert/modeling_wav2vec2_bert.py @@ -1013,10 +1013,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states, extract_features = self.feature_projection(input_features) diff --git a/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py b/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py index 15666f5fa2fc..2d41ebab2e41 100644 --- a/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py +++ b/src/transformers/models/wav2vec2_bert/modular_wav2vec2_bert.py @@ -723,7 +723,7 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py b/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py index dae00494bac3..74cdf50c1f80 100644 --- a/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py +++ b/src/transformers/models/wav2vec2_conformer/modeling_wav2vec2_conformer.py @@ -1173,10 +1173,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/wavlm/modeling_wavlm.py b/src/transformers/models/wavlm/modeling_wavlm.py index 7e2f1d757072..ff80f700d4d8 100755 --- a/src/transformers/models/wavlm/modeling_wavlm.py +++ b/src/transformers/models/wavlm/modeling_wavlm.py @@ -1084,10 +1084,8 @@ def forward( Indices to mask extracted features for contrastive loss. When in training mode, model learns to predict masked extracted features in *config.proj_codevector_dim* space. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict extract_features = self.feature_extractor(input_values) diff --git a/src/transformers/models/whisper/modeling_flax_whisper.py b/src/transformers/models/whisper/modeling_flax_whisper.py index 8cb98f2385d0..bcc25bdb9443 100644 --- a/src/transformers/models/whisper/modeling_flax_whisper.py +++ b/src/transformers/models/whisper/modeling_flax_whisper.py @@ -991,9 +991,7 @@ def encode( >>> encoder_outputs = model.encode(input_features=input_features) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # Handle any PRNG if needed @@ -1058,9 +1056,7 @@ def decode( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1148,9 +1144,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare decoder inputs @@ -1309,9 +1303,7 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1559,9 +1551,7 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if encoder_outputs is None: @@ -1636,9 +1626,7 @@ def __call__( **kwargs, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # Handle any PRNG if needed diff --git a/src/transformers/models/whisper/modeling_tf_whisper.py b/src/transformers/models/whisper/modeling_tf_whisper.py index f5ca846b81e3..caf1eca76260 100644 --- a/src/transformers/models/whisper/modeling_tf_whisper.py +++ b/src/transformers/models/whisper/modeling_tf_whisper.py @@ -749,9 +749,7 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 layers can't use channels first format when running on CPU. @@ -968,9 +966,7 @@ def call( Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1156,9 +1152,7 @@ def call( [1, 2, 512] ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/whisper/modeling_whisper.py b/src/transformers/models/whisper/modeling_whisper.py index 3e7f9c082ec3..75d118ce6883 100644 --- a/src/transformers/models/whisper/modeling_whisper.py +++ b/src/transformers/models/whisper/modeling_whisper.py @@ -677,10 +677,8 @@ def forward( f"Whisper expects the mel input features to be of length {expected_seq_length}, but found {input_features.shape[-1]}. Make sure to pad the input mel features to {expected_seq_length}." ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict inputs_embeds = nn.functional.gelu(self.conv1(input_features)) inputs_embeds = nn.functional.gelu(self.conv2(inputs_embeds)) @@ -866,10 +864,8 @@ def forward( Indices depicting the position of the input sequence tokens in the sequence. It is used to update the cache in the correct position and to infer the complete sequence length. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1168,10 +1164,8 @@ def forward( >>> list(last_hidden_state.shape) [1, 2, 512] ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1517,10 +1511,8 @@ def forward( >>> transcription ' Mr. Quilter is the apostle of the middle classes and we are glad to welcome his gospel.' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # If the user passed a tuple or `BaseModelOutput` for encoder_outputs, we extract only the hidden states @@ -1656,10 +1648,8 @@ def forward( 'Afrikaans' ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + if self.config.use_weighted_layer_sum: output_hidden_states = True elif output_hidden_states is None: diff --git a/src/transformers/models/x_clip/modeling_x_clip.py b/src/transformers/models/x_clip/modeling_x_clip.py index 7b300a9addfa..9df8d469a6d3 100644 --- a/src/transformers/models/x_clip/modeling_x_clip.py +++ b/src/transformers/models/x_clip/modeling_x_clip.py @@ -611,10 +611,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -673,10 +671,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is None: @@ -825,10 +821,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -892,10 +886,8 @@ def forward( interpolate_pos_encoding: bool = False, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(pixel_values, interpolate_pos_encoding=interpolate_pos_encoding) @@ -1234,10 +1226,8 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use X_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( @@ -1337,10 +1327,8 @@ def get_video_features( >>> video_features = model.get_video_features(**inputs) ```""" # Use X_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_frames, num_channels, height, width = pixel_values.shape @@ -1466,10 +1454,8 @@ def forward( tensor([[1.9496e-04, 9.9960e-01, 2.0825e-04]]) ```""" # Use X_CLIP model's config for some fields (if specified) instead of those of vision & text components. - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_frames, num_channels, height, width = pixel_values.shape diff --git a/src/transformers/models/xglm/modeling_flax_xglm.py b/src/transformers/models/xglm/modeling_flax_xglm.py index 1366148d9a3d..44d7c273ac94 100644 --- a/src/transformers/models/xglm/modeling_flax_xglm.py +++ b/src/transformers/models/xglm/modeling_flax_xglm.py @@ -624,9 +624,7 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/xglm/modeling_tf_xglm.py b/src/transformers/models/xglm/modeling_tf_xglm.py index 4ca60b5e5ad6..e7c6614f72e7 100644 --- a/src/transformers/models/xglm/modeling_tf_xglm.py +++ b/src/transformers/models/xglm/modeling_tf_xglm.py @@ -535,9 +535,7 @@ def call( **kwargs: Any, ) -> Union[TFBaseModelOutputWithPastAndCrossAttentions, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/xglm/modeling_xglm.py b/src/transformers/models/xglm/modeling_xglm.py index 4d1891747f4b..277913028079 100755 --- a/src/transformers/models/xglm/modeling_xglm.py +++ b/src/transformers/models/xglm/modeling_xglm.py @@ -465,10 +465,8 @@ def forward( - 1 indicates the head is **not masked**, - 0 indicates the head is **masked**. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -677,10 +675,8 @@ def forward( (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/xlm/modeling_xlm.py b/src/transformers/models/xlm/modeling_xlm.py index 8d22f3ddf5ec..c692faea3537 100755 --- a/src/transformers/models/xlm/modeling_xlm.py +++ b/src/transformers/models/xlm/modeling_xlm.py @@ -815,10 +815,8 @@ def forward( The dictionary object will be modified in-place during the forward pass to add newly computed hidden-states. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None: diff --git a/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py b/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py index bdbc06620a1b..3f0d34e5c9b7 100644 --- a/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py +++ b/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py @@ -840,9 +840,7 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py b/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py index b435f951a5ab..328e7a5230b9 100644 --- a/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py +++ b/src/transformers/models/xlm_roberta/modeling_xlm_roberta.py @@ -757,10 +757,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py b/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py index d24c0f2724b5..f2530d8f7d72 100644 --- a/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py +++ b/src/transformers/models/xlm_roberta_xl/modeling_xlm_roberta_xl.py @@ -749,10 +749,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/xlnet/modeling_xlnet.py b/src/transformers/models/xlnet/modeling_xlnet.py index 6c6450c25e47..ad951fb660a6 100755 --- a/src/transformers/models/xlnet/modeling_xlnet.py +++ b/src/transformers/models/xlnet/modeling_xlnet.py @@ -1272,10 +1272,8 @@ def forward( states from previous forward passes to compute attention, which can significantly improve performance for sequential decoding tasks. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if "use_cache" in kwargs: diff --git a/src/transformers/models/xmod/modeling_xmod.py b/src/transformers/models/xmod/modeling_xmod.py index 3d3c3aa4241a..059d6fa78bf9 100644 --- a/src/transformers/models/xmod/modeling_xmod.py +++ b/src/transformers/models/xmod/modeling_xmod.py @@ -761,10 +761,8 @@ def forward( Indices of the language adapters that should be activated for each sample, respectively. Default: the index that corresponds to `self.config.default_language`. """ - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/yolos/modeling_yolos.py b/src/transformers/models/yolos/modeling_yolos.py index 5e7c7daccb64..5a3027704643 100755 --- a/src/transformers/models/yolos/modeling_yolos.py +++ b/src/transformers/models/yolos/modeling_yolos.py @@ -589,10 +589,8 @@ def forward( head_mask: Optional[torch.Tensor] = None, return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/yoso/modeling_yoso.py b/src/transformers/models/yoso/modeling_yoso.py index c00fb97f5189..840debe7d5e0 100644 --- a/src/transformers/models/yoso/modeling_yoso.py +++ b/src/transformers/models/yoso/modeling_yoso.py @@ -699,10 +699,8 @@ def forward( return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithCrossAttentions]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_ids is not None and inputs_embeds is not None: diff --git a/src/transformers/models/zamba/modeling_zamba.py b/src/transformers/models/zamba/modeling_zamba.py index ff69c4defcd2..c04b51a150ad 100644 --- a/src/transformers/models/zamba/modeling_zamba.py +++ b/src/transformers/models/zamba/modeling_zamba.py @@ -929,10 +929,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1131,11 +1129,9 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/zamba2/modeling_zamba2.py b/src/transformers/models/zamba2/modeling_zamba2.py index 25f64745537a..2008919fa190 100644 --- a/src/transformers/models/zamba2/modeling_zamba2.py +++ b/src/transformers/models/zamba2/modeling_zamba2.py @@ -1291,10 +1291,8 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1552,11 +1550,9 @@ def forward( "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) + return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/zamba2/modular_zamba2.py b/src/transformers/models/zamba2/modular_zamba2.py index 99217d1b80a8..d442b0a52cd3 100644 --- a/src/transformers/models/zamba2/modular_zamba2.py +++ b/src/transformers/models/zamba2/modular_zamba2.py @@ -1045,7 +1045,7 @@ def forward( return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPast]: - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/zoedepth/modeling_zoedepth.py b/src/transformers/models/zoedepth/modeling_zoedepth.py index f1ae92d2ba3f..ada2fc14775b 100644 --- a/src/transformers/models/zoedepth/modeling_zoedepth.py +++ b/src/transformers/models/zoedepth/modeling_zoedepth.py @@ -1313,10 +1313,8 @@ def forward( raise NotImplementedError("Training is not implemented yet") return_dict = return_dict if return_dict is not None else self.config.use_return_dict - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + + outputs = self.backbone.forward_with_filtered_kwargs( pixel_values, output_hidden_states=output_hidden_states, output_attentions=output_attentions diff --git a/src/transformers/trainer.py b/src/transformers/trainer.py index bd0affc9ad32..996799fdf549 100755 --- a/src/transformers/trainer.py +++ b/src/transformers/trainer.py @@ -3941,7 +3941,7 @@ def save_model(self, output_dir: Optional[str] = None, _internal_call: bool = Fa self.push_to_hub(commit_message="Model save") def _save_tpu(self, output_dir: Optional[str] = None): - output_dir = output_dir if output_dir is not None else self.args.output_dir + logger.info(f"Saving model checkpoint to {output_dir}") model = self.model @@ -4014,7 +4014,7 @@ def _save_tpu(self, output_dir: Optional[str] = None): def _save(self, output_dir: Optional[str] = None, state_dict=None): # If we are executing this function, we are the process zero, so we don't check for that. - output_dir = output_dir if output_dir is not None else self.args.output_dir + os.makedirs(output_dir, exist_ok=True) logger.info(f"Saving model checkpoint to {output_dir}") From ce6092585fc4c4ec4363c417245c907188c1d6b8 Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 05:43:15 +0800 Subject: [PATCH 38/51] updates --- .../modular-transformers/modeling_test_detr.py | 18 ++++++++++++------ 1 file changed, 12 insertions(+), 6 deletions(-) diff --git a/examples/modular-transformers/modeling_test_detr.py b/examples/modular-transformers/modeling_test_detr.py index 3c7004d0fef6..de1084de7277 100644 --- a/examples/modular-transformers/modeling_test_detr.py +++ b/examples/modular-transformers/modeling_test_detr.py @@ -984,8 +984,10 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = inputs_embeds @@ -1123,8 +1125,10 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~file_utils.ModelOutput`] instead of a plain tuple. """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if inputs_embeds is not None: @@ -1520,8 +1524,10 @@ def forward( >>> list(last_hidden_states.shape) [1, 300, 256] ```""" - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, num_channels, height, width = pixel_values.shape From f1672862c189a5ab12fff5564b960b35ad61986f Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 05:44:13 +0800 Subject: [PATCH 39/51] more updats --- .../models/albert/modeling_albert.py | 76 +++++-------------- .../models/albert/modeling_flax_albert.py | 4 +- 2 files changed, 20 insertions(+), 60 deletions(-) diff --git a/src/transformers/models/albert/modeling_albert.py b/src/transformers/models/albert/modeling_albert.py index 7ce7f5b2c546..de8aeb19457c 100755 --- a/src/transformers/models/albert/modeling_albert.py +++ b/src/transformers/models/albert/modeling_albert.py @@ -22,6 +22,7 @@ import torch from torch import nn from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss +from transformers.utils.generic import can_return_tuple, check_model_inputs from ...activations import ACT2FN from ...modeling_attn_mask_utils import _prepare_4d_attention_mask_for_sdpa @@ -300,7 +301,6 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: bool = False, ) -> Union[tuple[torch.Tensor], tuple[torch.Tensor, torch.Tensor]]: mixed_query_layer = self.query(hidden_states) mixed_key_layer = self.key(hidden_states) @@ -351,7 +351,7 @@ def forward( projected_context_layer = self.dense(context_layer) projected_context_layer_dropout = self.output_dropout(projected_context_layer) layernormed_context_layer = self.LayerNorm(hidden_states + projected_context_layer_dropout) - return (layernormed_context_layer, attention_probs) if output_attentions else (layernormed_context_layer,) + return layernormed_context_layer, attention_probs class AlbertSdpaAttention(AlbertAttention): @@ -365,18 +365,7 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: bool = False, ) -> Union[tuple[torch.Tensor], tuple[torch.Tensor, torch.Tensor]]: - if self.position_embedding_type != "absolute" or output_attentions: - logger.warning( - "AlbertSdpaAttention is used but `torch.nn.functional.scaled_dot_product_attention` does not support " - "non-absolute `position_embedding_type` or `output_attentions=True` . Falling back to " - "the eager attention implementation, but specifying the eager implementation will be required from " - "Transformers version v5.0.0 onwards. This warning can be removed using the argument " - '`attn_implementation="eager"` when loading the model.' - ) - return super().forward(hidden_states, attention_mask, output_attentions=output_attentions) - batch_size, seq_len, _ = hidden_states.size() query_layer = self.transpose_for_scores(self.query(hidden_states)) key_layer = self.transpose_for_scores(self.key(hidden_states)) @@ -433,10 +422,8 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: bool = False, - output_hidden_states: bool = False, ) -> tuple[torch.Tensor, torch.Tensor]: - attention_output = self.attention(hidden_states, attention_mask, head_mask, output_attentions) + attention_output = self.attention(hidden_states, attention_mask, head_mask) ffn_output = apply_chunking_to_forward( self.ff_chunk, @@ -446,7 +433,7 @@ def forward( ) hidden_states = self.full_layer_layer_norm(ffn_output + attention_output[0]) - return (hidden_states,) + attention_output[1:] # add attentions if we output them + return hidden_states def ff_chunk(self, attention_output: torch.Tensor) -> torch.Tensor: ffn_output = self.ffn(attention_output) @@ -466,28 +453,12 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: bool = False, - output_hidden_states: bool = False, ) -> tuple[Union[torch.Tensor, tuple[torch.Tensor]], ...]: - layer_hidden_states = () - layer_attentions = () - for layer_index, albert_layer in enumerate(self.albert_layers): - layer_output = albert_layer(hidden_states, attention_mask, head_mask[layer_index], output_attentions) + layer_output = albert_layer(hidden_states, attention_mask, head_mask[layer_index]) hidden_states = layer_output[0] - if output_attentions: - layer_attentions = layer_attentions + (layer_output[1],) - - if output_hidden_states: - layer_hidden_states = layer_hidden_states + (hidden_states,) - - outputs = (hidden_states,) - if output_hidden_states: - outputs = outputs + (layer_hidden_states,) - if output_attentions: - outputs = outputs + (layer_attentions,) - return outputs # last-layer hidden state, (layer hidden states), (layer attentions) + return hidden_states class AlbertTransformer(nn.Module): @@ -498,20 +469,17 @@ def __init__(self, config: AlbertConfig): self.embedding_hidden_mapping_in = nn.Linear(config.embedding_size, config.hidden_size) self.albert_layer_groups = nn.ModuleList([AlbertLayerGroup(config) for _ in range(config.num_hidden_groups)]) + @can_return_tuple + @check_model_inputs def forward( self, hidden_states: torch.Tensor, attention_mask: Optional[torch.FloatTensor] = None, head_mask: Optional[torch.FloatTensor] = None, - output_attentions: bool = False, - output_hidden_states: bool = False, - return_dict: bool = True, + **kwargs, ) -> Union[BaseModelOutput, tuple]: hidden_states = self.embedding_hidden_mapping_in(hidden_states) - all_hidden_states = (hidden_states,) if output_hidden_states else None - all_attentions = () if output_attentions else None - head_mask = [None] * self.config.num_hidden_layers if head_mask is None else head_mask for i in range(self.config.num_hidden_layers): @@ -525,22 +493,10 @@ def forward( hidden_states, attention_mask, head_mask[group_idx * layers_per_group : (group_idx + 1) * layers_per_group], - output_attentions, - output_hidden_states, ) hidden_states = layer_group_output[0] - if output_attentions: - all_attentions = all_attentions + layer_group_output[-1] - - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - - if not return_dict: - return tuple(v for v in [hidden_states, all_hidden_states, all_attentions] if v is not None) - return BaseModelOutput( - last_hidden_state=hidden_states, hidden_states=all_hidden_states, attentions=all_attentions - ) + return BaseModelOutput(last_hidden_state=hidden_states) @auto_docstring @@ -663,7 +619,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, head_mask: Optional[torch.FloatTensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[BaseModelOutputWithPooling, tuple]: if input_ids is not None and inputs_embeds is not None: raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") @@ -716,10 +672,6 @@ def forward( sequence_output = encoder_outputs[0] pooled_output = self.pooler_activation(self.pooler(sequence_output[:, 0])) if self.pooler is not None else None - - if not return_dict: - return (sequence_output, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPooling( last_hidden_state=sequence_output, pooler_output=pooled_output, @@ -755,6 +707,7 @@ def get_input_embeddings(self) -> nn.Embedding: return self.albert.embeddings.word_embeddings @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -890,6 +843,7 @@ def get_input_embeddings(self) -> nn.Embedding: return self.albert.embeddings.word_embeddings @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -981,6 +935,7 @@ def __init__(self, config: AlbertConfig): self.post_init() @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -1062,6 +1017,7 @@ def __init__(self, config: AlbertConfig): self.post_init() @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -1112,6 +1068,7 @@ def __init__(self, config: AlbertConfig): self.post_init() @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -1178,6 +1135,7 @@ def __init__(self, config: AlbertConfig): self.post_init() @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, diff --git a/src/transformers/models/albert/modeling_flax_albert.py b/src/transformers/models/albert/modeling_flax_albert.py index b4f1d245c2e3..f2f19cb27716 100644 --- a/src/transformers/models/albert/modeling_flax_albert.py +++ b/src/transformers/models/albert/modeling_flax_albert.py @@ -566,7 +566,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed From 2d71cb33ac5dd4ccfb7808e3ab9d42b8b2338ca6 Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 05:44:23 +0800 Subject: [PATCH 40/51] revert tf changes --- src/transformers/generation/tf_utils.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/src/transformers/generation/tf_utils.py b/src/transformers/generation/tf_utils.py index 2b180922f1ef..be51c9cd9f43 100644 --- a/src/transformers/generation/tf_utils.py +++ b/src/transformers/generation/tf_utils.py @@ -1612,7 +1612,7 @@ def greedy_search( eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - + output_scores = output_scores if output_scores is not None else self.generation_config.output_scores output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) @@ -1896,7 +1896,7 @@ def sample( eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - + output_scores = output_scores if output_scores is not None else self.generation_config.output_scores output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) @@ -2241,7 +2241,7 @@ def unflatten_beam_dim(tensor, num_beams, batch_axis=0): output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.generation_config.output_hidden_states ) - + output_scores = output_scores if output_scores is not None else self.generation_config.output_scores return_dict_in_generate = ( return_dict_in_generate if return_dict_in_generate is not None @@ -2772,7 +2772,7 @@ def gather_fn(tensor): eos_token_id = eos_token_id if eos_token_id is not None else self.generation_config.eos_token_id if isinstance(eos_token_id, int): eos_token_id = [eos_token_id] - + output_scores = output_scores if output_scores is not None else self.generation_config.output_scores output_attentions = ( output_attentions if output_attentions is not None else self.generation_config.output_attentions ) From f3341519e2497040e20fa10453b2307895530730 Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 10:21:08 +0800 Subject: [PATCH 41/51] more changes propagation --- .../modular-transformers/modeling_dummy.py | 10 +- .../modeling_dummy_bert.py | 6 +- .../modeling_multimodal1.py | 10 +- .../modeling_multimodal2.py | 12 +- .../modeling_my_new_model2.py | 6 +- .../modeling_new_task_model.py | 6 +- .../modular-transformers/modeling_roberta.py | 6 +- .../models/align/modeling_align.py | 84 +---- .../models/altclip/modeling_altclip.py | 316 ++---------------- 9 files changed, 87 insertions(+), 369 deletions(-) diff --git a/examples/modular-transformers/modeling_dummy.py b/examples/modular-transformers/modeling_dummy.py index cb17ab459b2a..0fe4ae497b49 100644 --- a/examples/modular-transformers/modeling_dummy.py +++ b/examples/modular-transformers/modeling_dummy.py @@ -477,10 +477,12 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **kwargs: Unpack[FlashAttentionKwargs], + **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -537,7 +539,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **kwargs, + **flash_attn_kwargs, ) hidden_states = layer_outputs[0] diff --git a/examples/modular-transformers/modeling_dummy_bert.py b/examples/modular-transformers/modeling_dummy_bert.py index fc05912c4260..8b2e8aed90bc 100644 --- a/examples/modular-transformers/modeling_dummy_bert.py +++ b/examples/modular-transformers/modeling_dummy_bert.py @@ -926,8 +926,10 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/examples/modular-transformers/modeling_multimodal1.py b/examples/modular-transformers/modeling_multimodal1.py index 12dc815f9fb9..ec1a3346c9ba 100644 --- a/examples/modular-transformers/modeling_multimodal1.py +++ b/examples/modular-transformers/modeling_multimodal1.py @@ -477,10 +477,12 @@ def forward( output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, - **kwargs: Unpack[FlashAttentionKwargs], + **flash_attn_kwargs: Unpack[FlashAttentionKwargs], ) -> BaseModelOutputWithPast: - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): @@ -537,7 +539,7 @@ def forward( use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, - **kwargs, + **flash_attn_kwargs, ) hidden_states = layer_outputs[0] diff --git a/examples/modular-transformers/modeling_multimodal2.py b/examples/modular-transformers/modeling_multimodal2.py index 798c917489db..69e7e454754c 100644 --- a/examples/modular-transformers/modeling_multimodal2.py +++ b/examples/modular-transformers/modeling_multimodal2.py @@ -332,8 +332,10 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) encoder_states = () if output_hidden_states else None all_attentions = () if output_attentions else None @@ -499,8 +501,10 @@ def forward( Returns: """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if pixel_values is None: raise ValueError("You have to specify pixel_values") diff --git a/examples/modular-transformers/modeling_my_new_model2.py b/examples/modular-transformers/modeling_my_new_model2.py index 2cd9bad12b68..d8e10885ef8a 100644 --- a/examples/modular-transformers/modeling_my_new_model2.py +++ b/examples/modular-transformers/modeling_my_new_model2.py @@ -477,8 +477,10 @@ def forward( cache_position: Optional[torch.LongTensor] = None, **kwargs, # NOOP kwarg for now ) -> BaseModelOutputWithPast: - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache if (input_ids is None) ^ (inputs_embeds is not None): diff --git a/examples/modular-transformers/modeling_new_task_model.py b/examples/modular-transformers/modeling_new_task_model.py index d7c1a48fa463..77e4efa172e8 100644 --- a/examples/modular-transformers/modeling_new_task_model.py +++ b/examples/modular-transformers/modeling_new_task_model.py @@ -291,8 +291,10 @@ def forward( if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict is_training = token_type_ids is not None and labels is not None diff --git a/examples/modular-transformers/modeling_roberta.py b/examples/modular-transformers/modeling_roberta.py index 0ac2fdd4e0e3..e1bd313a4243 100644 --- a/examples/modular-transformers/modeling_roberta.py +++ b/examples/modular-transformers/modeling_roberta.py @@ -929,8 +929,10 @@ def forward( If set to `True`, `past_key_values` key value states are returned and can be used to speed up decoding (see `past_key_values`). """ - - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if self.config.is_decoder: diff --git a/src/transformers/models/align/modeling_align.py b/src/transformers/models/align/modeling_align.py index 94c6e72521cb..03479d72c388 100644 --- a/src/transformers/models/align/modeling_align.py +++ b/src/transformers/models/align/modeling_align.py @@ -32,6 +32,7 @@ from ...modeling_utils import PreTrainedModel from ...pytorch_utils import apply_chunking_to_forward, find_pruneable_heads_and_indices, prune_linear_layer from ...utils import ModelOutput, auto_docstring, logging +from transformers.utils.generic import can_return_tuple, check_model_inputs from .configuration_align import AlignConfig, AlignTextConfig, AlignVisionConfig @@ -507,10 +508,8 @@ def forward( ) -> BaseModelOutputWithPoolingAndNoAttention: for block in self.blocks: hidden_states = block(hidden_states) - return BaseModelOutputWithNoAttention( last_hidden_state=hidden_states, - hidden_states=all_hidden_states, ) @@ -764,7 +763,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: self_outputs = self.self( hidden_states, @@ -773,7 +771,6 @@ def forward( encoder_hidden_states, encoder_attention_mask, past_key_value, - output_attentions, ) attention_output = self.output(self_outputs[0], hidden_states) outputs = (attention_output,) + self_outputs[1:] # add attentions if we output them @@ -844,11 +841,6 @@ def forward( head_mask, past_key_value=self_attn_past_key_value, ) - attention_output = self_attention_outputs[0] - if self.is_decoder: - outputs = self_attention_outputs[1:-1] - else: - outputs = self_attention_outputs[1:] # add self attentions if we output attention weights if self.is_decoder and encoder_hidden_states is not None: if not hasattr(self, "crossattention"): @@ -859,18 +851,18 @@ def forward( cross_attn_past_key_value = past_key_value[-2:] if past_key_value is not None else None cross_attention_outputs = self.crossattention( - attention_output, attention_mask, head_mask, encoder_hidden_states, encoder_attention_mask, cross_attn_past_key_value, ) - attention_output = cross_attention_outputs[0] - outputs = outputs + cross_attention_outputs[1:-1] # add cross attentions if we output attention weights layer_output = apply_chunking_to_forward( - self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output + self.feed_forward_chunk, + self.chunk_size_feed_forward, + self.seq_len_dim, + (self_attention_outputs, cross_attention_outputs), ) return layer_output @@ -897,7 +889,7 @@ def forward( encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, **kwargs, - ) -> Union[BaseModelOutputWithPastAndCrossAttentions]: + ) -> BaseModelOutputWithPastAndCrossAttentions: for i, layer_module in enumerate(self.layer): layer_head_mask = head_mask[i] if head_mask is not None else None past_key_value = past_key_values[i] if past_key_values is not None else None @@ -991,6 +983,7 @@ def set_input_embeddings(self, value): self.embeddings.word_embeddings = value @auto_docstring + @check_model_inputs def forward( self, input_ids: Optional[torch.Tensor] = None, @@ -1000,7 +993,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, **kwargs, - ) -> Union[tuple, BaseModelOutputWithPoolingAndCrossAttentions]: + ) -> BaseModelOutputWithPoolingAndCrossAttentions: r""" Examples: @@ -1016,19 +1009,10 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled (EOS token) states ```""" - if input_ids is not None and inputs_embeds is not None: - raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") - elif input_ids is not None: - self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask) - input_shape = input_ids.size() - elif inputs_embeds is not None: - input_shape = inputs_embeds.size()[:-1] - else: - raise ValueError("You have to specify either input_ids or inputs_embeds") - - batch_size, seq_length = input_shape + input_shape = input_ids.shape if input_ids is not None else inputs_embeds.shape[:-1] device = input_ids.device if input_ids is not None else inputs_embeds.device + batch_size, seq_length = input_shape if attention_mask is None: attention_mask = torch.ones(((batch_size, seq_length)), device=device) @@ -1098,12 +1082,10 @@ def __init__(self, config: AlignVisionConfig): def get_input_embeddings(self) -> nn.Module: return self.vision_model.embeddings.convolution + @can_return_tuple @auto_docstring def forward( - self, - pixel_values: Optional[torch.FloatTensor] = None, - output_hidden_states: Optional[bool] = None, - return_dict: Optional[bool] = None, + self, pixel_values: Optional[torch.FloatTensor] = None, **kwargs ) -> Union[tuple, BaseModelOutputWithPoolingAndNoAttention]: r""" Examples: @@ -1125,8 +1107,6 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled CLS states ```""" - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -1134,8 +1114,6 @@ def forward( embedding_output = self.embeddings(pixel_values) encoder_outputs = self.encoder( embedding_output, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) # Apply pooling last_hidden_state = encoder_outputs[0] @@ -1143,13 +1121,9 @@ def forward( # Reshape (batch_size, projection_dim, 1 , 1) -> (batch_size, projection_dim) pooled_output = pooled_output.reshape(pooled_output.shape[:2]) - if not return_dict: - return (last_hidden_state, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPoolingAndNoAttention( last_hidden_state=last_hidden_state, pooler_output=pooled_output, - hidden_states=encoder_outputs.hidden_states, ) @@ -1188,6 +1162,7 @@ def __init__(self, config: AlignConfig): self.post_init() @auto_docstring + @can_return_tuple def get_text_features( self, input_ids: Optional[torch.Tensor] = None, @@ -1196,7 +1171,7 @@ def get_text_features( position_ids: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, + **kwargs, ) -> torch.FloatTensor: r""" Returns: @@ -1214,11 +1189,6 @@ def get_text_features( >>> inputs = tokenizer(["a photo of a cat", "a photo of a dog"], padding=True, return_tensors="pt") >>> text_features = model.get_text_features(**inputs) ```""" - # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - text_outputs = self.text_model( input_ids=input_ids, attention_mask=attention_mask, @@ -1226,9 +1196,7 @@ def get_text_features( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, + **kwargs, ) last_hidden_state = text_outputs[0][:, 0, :] @@ -1240,8 +1208,6 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - output_hidden_states: Optional[bool] = None, - return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" Returns: @@ -1266,13 +1232,9 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( pixel_values=pixel_values, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) image_features = vision_outputs[1] # pooled_output @@ -1290,7 +1252,7 @@ def forward( head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[tuple, AlignOutput]: r""" return_loss (`bool`, *optional*): @@ -1318,14 +1280,8 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use ALIGN model's config for some fields (if specified) instead of those of vision & text components. - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - vision_outputs = self.vision_model( pixel_values=pixel_values, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) text_outputs = self.text_model( @@ -1335,9 +1291,6 @@ def forward( position_ids=position_ids, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) image_embeds = vision_outputs[1] @@ -1355,11 +1308,6 @@ def forward( loss = None if return_loss: loss = align_loss(logits_per_text) - - if not return_dict: - output = (logits_per_image, logits_per_text, text_embeds, image_embeds, text_outputs, vision_outputs) - return ((loss,) + output) if loss is not None else output - return AlignOutput( loss=loss, logits_per_image=logits_per_image, diff --git a/src/transformers/models/altclip/modeling_altclip.py b/src/transformers/models/altclip/modeling_altclip.py index 899be3e1d58c..84263654bcad 100755 --- a/src/transformers/models/altclip/modeling_altclip.py +++ b/src/transformers/models/altclip/modeling_altclip.py @@ -20,7 +20,8 @@ import torch import torch.nn as nn -import torch.utils.checkpoint + +from transformers.utils.generic import can_return_tuple from ...activations import ACT2FN from ...modeling_outputs import ( @@ -30,7 +31,7 @@ BaseModelOutputWithPoolingAndCrossAttentions, BaseModelOutputWithPoolingAndProjection, ) -from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel +from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, GradientCheckpointingLayer, PreTrainedModel from ...pytorch_utils import apply_chunking_to_forward, find_pruneable_heads_and_indices, prune_linear_layer from ...utils import ModelOutput, auto_docstring, logging, torch_int from .configuration_altclip import AltCLIPConfig, AltCLIPTextConfig, AltCLIPVisionConfig @@ -220,7 +221,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: mixed_query_layer = self.query(hidden_states) @@ -307,11 +307,7 @@ def forward( new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,) context_layer = context_layer.view(new_context_layer_shape) - outputs = (context_layer, attention_probs) if output_attentions else (context_layer,) - - if self.is_decoder: - outputs = outputs + (past_key_value,) - return outputs + return context_layer, attention_probs # Copied from transformers.models.roberta.modeling_roberta.RobertaSelfOutput @@ -370,7 +366,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: self_outputs = self.self( hidden_states, @@ -379,7 +374,6 @@ def forward( encoder_hidden_states, encoder_attention_mask, past_key_value, - output_attentions, ) attention_output = self.output(self_outputs[0], hidden_states) outputs = (attention_output,) + self_outputs[1:] # add attentions if we output them @@ -441,7 +435,6 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_value: Optional[tuple[tuple[torch.FloatTensor]]] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor]: # decoder uni-directional self-attention cached key/values tuple is at positions 1,2 self_attn_past_key_value = past_key_value[:2] if past_key_value is not None else None @@ -449,7 +442,6 @@ def forward( hidden_states, attention_mask, head_mask, - output_attentions=output_attentions, past_key_value=self_attn_past_key_value, ) attention_output = self_attention_outputs[0] @@ -478,24 +470,17 @@ def forward( encoder_hidden_states, encoder_attention_mask, cross_attn_past_key_value, - output_attentions, ) attention_output = cross_attention_outputs[0] outputs = outputs + cross_attention_outputs[1:-1] # add cross attentions if we output attention weights # add cross-attn cache to positions 3,4 of present_key_value tuple cross_attn_present_key_value = cross_attention_outputs[-1] - present_key_value = present_key_value + cross_attn_present_key_value layer_output = apply_chunking_to_forward( self.feed_forward_chunk, self.chunk_size_feed_forward, self.seq_len_dim, attention_output ) outputs = (layer_output,) + outputs - - # if decoder, return the attn key/values as the last output - if self.is_decoder: - outputs = outputs + (present_key_value,) - return outputs def feed_forward_chunk(self, attention_output): @@ -512,6 +497,7 @@ def __init__(self, config): self.layer = nn.ModuleList([AltRobertaLayer(config) for _ in range(config.num_hidden_layers)]) self.gradient_checkpointing = False + @can_return_tuple def forward( self, hidden_states: torch.Tensor, @@ -520,82 +506,25 @@ def forward( encoder_hidden_states: Optional[torch.FloatTensor] = None, encoder_attention_mask: Optional[torch.FloatTensor] = None, past_key_values: Optional[tuple[tuple[torch.FloatTensor]]] = None, - use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = False, - output_hidden_states: Optional[bool] = False, - return_dict: Optional[bool] = True, - ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: - all_hidden_states = () if output_hidden_states else None - all_self_attentions = () if output_attentions else None - all_cross_attentions = () if output_attentions and self.config.add_cross_attention else None - - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - - next_decoder_cache = () if use_cache else None + **kwargs + ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPastAndCrossAttentions]: + for i, layer_module in enumerate(self.layer): - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - layer_head_mask = head_mask[i] if head_mask is not None else None past_key_value = past_key_values[i] if past_key_values is not None else None - - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - layer_module.__call__, - hidden_states, - attention_mask, - layer_head_mask, - encoder_hidden_states, - encoder_attention_mask, - past_key_value, - output_attentions, - ) - else: - layer_outputs = layer_module( - hidden_states, - attention_mask, - layer_head_mask, - encoder_hidden_states, - encoder_attention_mask, - past_key_value, - output_attentions, - ) + layer_outputs = layer_module( + hidden_states, + attention_mask, + layer_head_mask, + encoder_hidden_states, + encoder_attention_mask, + past_key_value, + ) hidden_states = layer_outputs[0] - if use_cache: - next_decoder_cache += (layer_outputs[-1],) - if output_attentions: - all_self_attentions = all_self_attentions + (layer_outputs[1],) - if self.config.add_cross_attention: - all_cross_attentions = all_cross_attentions + (layer_outputs[2],) - - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - - if not return_dict: - return tuple( - v - for v in [ - hidden_states, - next_decoder_cache, - all_hidden_states, - all_self_attentions, - all_cross_attentions, - ] - if v is not None - ) - return BaseModelOutputWithPastAndCrossAttentions( + return BaseModelOutputWithPastAndCrossAttentions( last_hidden_state=hidden_states, - past_key_values=next_decoder_cache, - hidden_states=all_hidden_states, - attentions=all_self_attentions, - cross_attentions=all_cross_attentions, - ) + ) # Copied from transformers.models.roberta.modeling_roberta.RobertaPooler @@ -665,7 +594,6 @@ def forward( hidden_states: torch.Tensor, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = False, ) -> tuple[torch.Tensor, Optional[torch.Tensor]]: """Input shape: Batch x Time x Channel""" @@ -690,13 +618,7 @@ def forward( attention_interface: Callable = eager_attention_forward if self.config._attn_implementation != "eager": - if self.config._attn_implementation == "sdpa" and output_attentions: - logger.warning_once( - "`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to " - 'eager attention. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.' - ) - else: - attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] + attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] attn_output, attn_weights = attention_interface( self, @@ -711,8 +633,6 @@ def forward( attn_output = attn_output.reshape(batch_size, seq_length, embed_dim).contiguous() attn_output = self.out_proj(attn_output) - if not output_attentions: - attn_weights = None return attn_output, attn_weights @@ -732,7 +652,7 @@ def forward(self, hidden_states: torch.Tensor) -> torch.Tensor: return hidden_states -class AltCLIPEncoderLayer(nn.Module): +class AltCLIPEncoderLayer(GradientCheckpointingLayer): def __init__(self, config: AltCLIPConfig): super().__init__() self.embed_dim = config.hidden_size @@ -746,18 +666,7 @@ def forward( hidden_states: torch.Tensor, attention_mask: torch.Tensor, causal_attention_mask: torch.Tensor, - output_attentions: Optional[bool] = False, ) -> tuple[torch.FloatTensor]: - """ - Args: - hidden_states (`torch.FloatTensor`): input to the layer of shape `(batch, seq_len, embed_dim)` - attention_mask (`torch.FloatTensor`): attention mask of size - `(batch, 1, tgt_len, src_len)` where padding elements are indicated by very large negative values. - `(config.encoder_attention_heads,)`. - output_attentions (`bool`, *optional*): - Whether or not to return the attentions tensors of all attention layers. See `attentions` under - returned tensors for more detail. - """ residual = hidden_states hidden_states = self.layer_norm1(hidden_states) @@ -765,7 +674,6 @@ def forward( hidden_states=hidden_states, attention_mask=attention_mask, causal_attention_mask=causal_attention_mask, - output_attentions=output_attentions, ) hidden_states = residual + hidden_states @@ -773,14 +681,7 @@ def forward( hidden_states = self.layer_norm2(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - - outputs = (hidden_states,) - - if output_attentions: - outputs += (attn_weights,) - - return outputs - + return hidden_states, attn_weights class AltCLIPEncoder(nn.Module): """ @@ -802,78 +703,18 @@ def forward( inputs_embeds, attention_mask: Optional[torch.Tensor] = None, causal_attention_mask: Optional[torch.Tensor] = None, - - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: - r""" - Args: - inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`): - Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. - This is useful if you want more control over how to convert `input_ids` indices into associated vectors - than the model's internal embedding lookup matrix. - attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*): - Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`: - - - 1 for tokens that are **not masked**, - - 0 for tokens that are **masked**. - - [What are attention masks?](../glossary#attention-mask) - causal_attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*): - Causal mask for the text model. Mask values selected in `[0, 1]`: - - - 1 for tokens that are **not masked**, - - 0 for tokens that are **masked**. - - [What are attention masks?](../glossary#attention-mask) - output_attentions (`bool`, *optional*): - Whether or not to return the attentions tensors of all attention layers. See `attentions` under - returned tensors for more detail. - output_hidden_states (`bool`, *optional*): - Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors - for more detail. - return_dict (`bool`, *optional*): - Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. - """ - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - - encoder_states = () if output_hidden_states else None - all_attentions = () if output_attentions else None - hidden_states = inputs_embeds for idx, encoder_layer in enumerate(self.layers): - if output_hidden_states: - encoder_states = encoder_states + (hidden_states,) - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - encoder_layer.__call__, - hidden_states, - attention_mask, - causal_attention_mask, - output_attentions, - ) - else: - layer_outputs = encoder_layer( - hidden_states, - attention_mask, - causal_attention_mask, - output_attentions=output_attentions, - ) + layer_outputs = encoder_layer( + hidden_states, + attention_mask, + causal_attention_mask, + ) hidden_states = layer_outputs[0] - if output_attentions: - all_attentions = all_attentions + (layer_outputs[1],) - - if output_hidden_states: - encoder_states = encoder_states + (hidden_states,) - - if not return_dict: - return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None) - return BaseModelOutput( - last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions - ) + return BaseModelOutput(last_hidden_state=hidden_states) # Copied from transformers.models.clip.modeling_clip.CLIPVisionEmbeddings with CLIP->AltCLIP @@ -1025,16 +866,12 @@ def __init__(self, config: AltCLIPVisionConfig): self.post_layernorm = nn.LayerNorm(embed_dim, eps=config.layer_norm_eps) @auto_docstring + @can_return_tuple def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - - return_dict: Optional[bool] = None, interpolate_pos_encoding: Optional[bool] = False, ) -> Union[tuple, BaseModelOutputWithPooling]: - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: raise ValueError("You have to specify pixel_values") @@ -1044,18 +881,12 @@ def forward( encoder_outputs = self.encoder( inputs_embeds=hidden_states, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, - ) + ) last_hidden_state = encoder_outputs[0] pooled_output = last_hidden_state[:, 0, :] pooled_output = self.post_layernorm(pooled_output) - if not return_dict: - return (last_hidden_state, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPooling( last_hidden_state=last_hidden_state, pooler_output=pooled_output, @@ -1081,9 +912,7 @@ def get_input_embeddings(self) -> nn.Module: def forward( self, pixel_values: Optional[torch.FloatTensor] = None, - interpolate_pos_encoding: bool = False, - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" Examples: @@ -1105,14 +934,10 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state >>> pooled_output = outputs.pooler_output # pooled CLS states ```""" - return_dict = return_dict if return_dict is not None else self.config.use_return_dict return self.vision_model( pixel_values=pixel_values, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - interpolate_pos_encoding=interpolate_pos_encoding, - return_dict=return_dict, + interpolate_pos_encoding=interpolate_pos_encoding, ) @@ -1165,6 +990,7 @@ class PreTrainedModel self.encoder.layer[layer].attention.prune_heads(heads) @auto_docstring + @check_model_inputs # Copied from transformers.models.clap.modeling_clap.ClapTextModel.forward def forward( self, @@ -1178,28 +1004,7 @@ def forward( encoder_attention_mask: Optional[torch.Tensor] = None, past_key_values: Optional[list[torch.FloatTensor]] = None, use_cache: Optional[bool] = None, - - return_dict: Optional[bool] = None, ) -> Union[tuple[torch.Tensor], BaseModelOutputWithPoolingAndCrossAttentions]: - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - - if self.config.is_decoder: - use_cache = use_cache if use_cache is not None else self.config.use_cache - else: - use_cache = False - - if input_ids is not None and inputs_embeds is not None: - raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") - elif input_ids is not None: - self.warn_if_padding_and_no_attention_mask(input_ids, attention_mask) - input_shape = input_ids.size() - elif inputs_embeds is not None: - input_shape = inputs_embeds.size()[:-1] - else: - raise ValueError("You have to specify either input_ids or inputs_embeds") - batch_size, seq_length = input_shape device = input_ids.device if input_ids is not None else inputs_embeds.device @@ -1254,23 +1059,13 @@ def forward( encoder_attention_mask=encoder_extended_attention_mask, past_key_values=past_key_values, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, - ) + ) sequence_output = encoder_outputs[0] pooled_output = self.pooler(sequence_output) if self.pooler is not None else None - if not return_dict: - return (sequence_output, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPoolingAndCrossAttentions( last_hidden_state=sequence_output, pooler_output=pooled_output, - past_key_values=encoder_outputs.past_key_values, - hidden_states=encoder_outputs.hidden_states, - attentions=encoder_outputs.attentions, - cross_attentions=encoder_outputs.cross_attentions, ) @@ -1294,6 +1089,7 @@ def resize_token_embeddings(self, new_num_tokens: Optional[int] = None) -> nn.Em return super().resize_token_embeddings(new_num_tokens) @auto_docstring + @can_return_tuple def forward( self, input_ids: Optional[torch.Tensor] = None, @@ -1304,9 +1100,6 @@ def forward( inputs_embeds: Optional[torch.Tensor] = None, encoder_hidden_states: Optional[torch.Tensor] = None, encoder_attention_mask: Optional[torch.Tensor] = None, - output_attentions: Optional[bool] = None, - return_dict: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPoolingAndProjection]: r""" Examples: @@ -1326,7 +1119,6 @@ def forward( >>> pooled_output = outputs.pooler_output # pooled CLS states ```""" - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.roberta( input_ids=input_ids, @@ -1337,9 +1129,6 @@ def forward( inputs_embeds=inputs_embeds, encoder_hidden_states=encoder_hidden_states, encoder_attention_mask=encoder_attention_mask, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) # last module outputs @@ -1351,15 +1140,9 @@ def forward( # pooler projection_state = self.transformation(sequence_output) pooler_output = projection_state[:, 0] - - if not return_dict: - return (projection_state, pooler_output) + outputs[2:4] - return BaseModelOutputWithPoolingAndProjection( last_hidden_state=projection_state, pooler_output=pooler_output, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) @@ -1404,8 +1187,6 @@ def get_text_features( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.Tensor] = None, token_type_ids=None, - - return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" Returns: @@ -1423,18 +1204,12 @@ def get_text_features( >>> text_features = model.get_text_features(**inputs) ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict text_outputs = self.text_model( input_ids=input_ids, attention_mask=attention_mask, position_ids=position_ids, token_type_ids=token_type_ids, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) pooled_output = text_outputs[1] text_features = self.text_projection(pooled_output) @@ -1445,9 +1220,7 @@ def get_text_features( def get_image_features( self, pixel_values: Optional[torch.FloatTensor] = None, - interpolate_pos_encoding: bool = False, - return_dict: Optional[bool] = None, ) -> torch.FloatTensor: r""" Returns: @@ -1469,16 +1242,11 @@ def get_image_features( >>> image_features = model.get_image_features(**inputs) ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict + vision_outputs = self.vision_model( pixel_values=pixel_values, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, interpolate_pos_encoding=interpolate_pos_encoding, - return_dict=return_dict, ) pooled_output = vision_outputs[1] # pooled_output @@ -1495,9 +1263,7 @@ def forward( position_ids: Optional[torch.LongTensor] = None, token_type_ids: Optional[torch.Tensor] = None, return_loss: Optional[bool] = None, - interpolate_pos_encoding: bool = False, - return_dict: Optional[bool] = None, ) -> Union[tuple, AltCLIPOutput]: r""" return_loss (`bool`, *optional*): @@ -1522,26 +1288,18 @@ def forward( >>> probs = logits_per_image.softmax(dim=1) # we can take the softmax to get the label probabilities ```""" # Use AltCLIP model's config for some fields (if specified) instead of those of vision & text components. - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict + text_outputs = self.text_model( input_ids=input_ids, attention_mask=attention_mask, token_type_ids=token_type_ids, position_ids=position_ids, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) vision_outputs = self.vision_model( pixel_values=pixel_values, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, interpolate_pos_encoding=interpolate_pos_encoding, - return_dict=return_dict, ) image_embeds = vision_outputs[1] @@ -1563,10 +1321,6 @@ def forward( if return_loss: loss = clip_loss(logits_per_text) - if not return_dict: - output = (logits_per_image, logits_per_text, text_embeds, image_embeds, text_outputs, vision_outputs) - return ((loss,) + output) if loss is not None else output - return AltCLIPOutput( loss=loss, logits_per_image=logits_per_image, From 54242e21cf6ec6a5ed8b6693e67fc713cbf90143 Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 10:33:22 +0800 Subject: [PATCH 42/51] revert changes to the cursted libraries --- src/transformers/models/aria/modular_aria.py | 19 --------------- .../models/bart/modeling_flax_bart.py | 20 ++++++++++++---- .../models/beit/modeling_flax_beit.py | 4 +++- .../models/bert/modeling_flax_bert.py | 4 +++- .../models/big_bird/modeling_flax_big_bird.py | 8 +++++-- .../blenderbot/modeling_flax_blenderbot.py | 16 +++++++++---- .../blenderbot/modeling_tf_blenderbot.py | 4 +++- .../modeling_flax_blenderbot_small.py | 16 +++++++++---- .../modeling_tf_blenderbot_small.py | 4 +++- .../models/blip/modeling_tf_blip.py | 12 +++++++--- .../models/blip/modeling_tf_blip_text.py | 4 +++- .../models/bloom/modeling_flax_bloom.py | 4 +++- .../models/clip/modeling_flax_clip.py | 20 ++++++++++++---- .../models/convnext/modeling_tf_convnext.py | 12 +++++++--- .../convnextv2/modeling_tf_convnextv2.py | 12 +++++++--- .../data2vec/modeling_tf_data2vec_vision.py | 8 +++++-- .../models/deit/modeling_tf_deit.py | 4 +++- .../modeling_tf_efficientformer.py | 4 +++- .../models/dinov2/modeling_flax_dinov2.py | 4 +++- .../distilbert/modeling_flax_distilbert.py | 8 +++++-- .../models/electra/modeling_flax_electra.py | 4 +++- .../modeling_flax_encoder_decoder.py | 12 +++++++--- .../models/gemma/modeling_flax_gemma.py | 4 +++- .../models/gpt2/modeling_flax_gpt2.py | 4 +++- .../models/gpt_neo/modeling_flax_gpt_neo.py | 4 +++- .../models/gptj/modeling_flax_gptj.py | 4 +++- .../models/idefics/modeling_tf_idefics.py | 8 +++++-- src/transformers/models/idefics/vision_tf.py | 4 ++-- .../layoutlmv3/modeling_tf_layoutlmv3.py | 4 +++- .../models/llama/modeling_flax_llama.py | 4 +++- .../models/longt5/modeling_flax_longt5.py | 16 +++++++++---- .../models/marian/modeling_flax_marian.py | 16 +++++++++---- .../models/marian/modeling_tf_marian.py | 4 +++- .../models/mbart/modeling_flax_mbart.py | 16 +++++++++---- .../models/mbart/modeling_tf_mbart.py | 4 +++- .../models/mistral/modeling_flax_mistral.py | 4 +++- .../models/mobilevit/modeling_tf_mobilevit.py | 8 +++++-- .../models/opt/modeling_flax_opt.py | 4 +++- .../models/opt/modeling_tf_opt.py | 16 +++++++++---- .../models/pegasus/modeling_flax_pegasus.py | 16 +++++++++---- .../models/pegasus/modeling_tf_pegasus.py | 4 +++- .../models/regnet/modeling_flax_regnet.py | 8 +++++-- .../models/regnet/modeling_tf_regnet.py | 12 +++++++--- .../models/resnet/modeling_flax_resnet.py | 8 +++++-- .../models/resnet/modeling_tf_resnet.py | 8 +++++-- .../models/roberta/modeling_flax_roberta.py | 4 +++- .../modeling_flax_roberta_prelayernorm.py | 4 +++- .../models/roformer/modeling_flax_roformer.py | 4 +++- .../models/sam/modeling_tf_sam.py | 12 +++++++--- .../models/segformer/modeling_tf_segformer.py | 8 +++++-- .../modeling_flax_speech_encoder_decoder.py | 12 +++++++--- .../modeling_tf_speech_to_text.py | 4 +++- .../swiftformer/modeling_tf_swiftformer.py | 8 +++++-- .../models/swin/modeling_tf_swin.py | 8 +++++-- .../models/t5/modeling_flax_t5.py | 20 ++++++++++++---- .../models/whisper/modeling_flax_whisper.py | 24 ++++++++++++++----- .../models/whisper/modeling_tf_whisper.py | 12 +++++++--- .../models/xglm/modeling_flax_xglm.py | 4 +++- .../models/xglm/modeling_tf_xglm.py | 4 +++- .../xlm_roberta/modeling_flax_xlm_roberta.py | 4 +++- 60 files changed, 371 insertions(+), 144 deletions(-) diff --git a/src/transformers/models/aria/modular_aria.py b/src/transformers/models/aria/modular_aria.py index 0335deab2f70..b263695df425 100644 --- a/src/transformers/models/aria/modular_aria.py +++ b/src/transformers/models/aria/modular_aria.py @@ -1425,17 +1425,9 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AriaModelOutputWithPast]: - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if inputs_embeds is None: inputs_embeds = self.get_input_embeddings()(input_ids) @@ -1471,8 +1463,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, return_dict=True, cache_position=cache_position, **kwargs, @@ -1521,7 +1511,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, logits_to_keep: Union[int, torch.Tensor] = 0, cache_position: Optional[torch.LongTensor] = None, @@ -1587,11 +1576,6 @@ def forward( >>> print(generated_texts[1]) Assistant: The bridge is in San Francisco. ```""" - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( input_ids=input_ids, @@ -1602,9 +1586,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, **kwargs, ) diff --git a/src/transformers/models/bart/modeling_flax_bart.py b/src/transformers/models/bart/modeling_flax_bart.py index 07eee3deee08..818254f3bfa1 100644 --- a/src/transformers/models/bart/modeling_flax_bart.py +++ b/src/transformers/models/bart/modeling_flax_bart.py @@ -1026,7 +1026,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1097,7 +1099,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1186,7 +1190,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1360,7 +1366,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1804,7 +1812,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/beit/modeling_flax_beit.py b/src/transformers/models/beit/modeling_flax_beit.py index 0cd7e99d1a69..c80deace6b39 100644 --- a/src/transformers/models/beit/modeling_flax_beit.py +++ b/src/transformers/models/beit/modeling_flax_beit.py @@ -642,7 +642,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/bert/modeling_flax_bert.py b/src/transformers/models/bert/modeling_flax_bert.py index 7ea01757a97d..37828642eb4e 100644 --- a/src/transformers/models/bert/modeling_flax_bert.py +++ b/src/transformers/models/bert/modeling_flax_bert.py @@ -873,7 +873,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/big_bird/modeling_flax_big_bird.py b/src/transformers/models/big_bird/modeling_flax_big_bird.py index 4c782e148e16..11dcb30f3d47 100644 --- a/src/transformers/models/big_bird/modeling_flax_big_bird.py +++ b/src/transformers/models/big_bird/modeling_flax_big_bird.py @@ -1735,7 +1735,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed @@ -2449,7 +2451,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: diff --git a/src/transformers/models/blenderbot/modeling_flax_blenderbot.py b/src/transformers/models/blenderbot/modeling_flax_blenderbot.py index 8f71a00f2faf..8b147211881b 100644 --- a/src/transformers/models/blenderbot/modeling_flax_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_flax_blenderbot.py @@ -999,7 +999,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1072,7 +1074,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1161,7 +1165,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1336,7 +1342,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/blenderbot/modeling_tf_blenderbot.py b/src/transformers/models/blenderbot/modeling_tf_blenderbot.py index 4cd29be2bfbb..86e576c98280 100644 --- a/src/transformers/models/blenderbot/modeling_tf_blenderbot.py +++ b/src/transformers/models/blenderbot/modeling_tf_blenderbot.py @@ -1139,7 +1139,9 @@ def call( training=False, **kwargs, ): - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py index 97044b3439e3..ac30320bbdb4 100644 --- a/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_flax_blenderbot_small.py @@ -996,7 +996,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1069,7 +1071,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1157,7 +1161,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1333,7 +1339,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py b/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py index aac28c893586..459ef438c53c 100644 --- a/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py +++ b/src/transformers/models/blenderbot_small/modeling_tf_blenderbot_small.py @@ -1140,7 +1140,9 @@ def call( training=False, **kwargs, ): - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/blip/modeling_tf_blip.py b/src/transformers/models/blip/modeling_tf_blip.py index bec4b5a2c384..a0bf43892d0e 100644 --- a/src/transformers/models/blip/modeling_tf_blip.py +++ b/src/transformers/models/blip/modeling_tf_blip.py @@ -650,7 +650,9 @@ def call( Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict encoder_states = () if output_hidden_states else None @@ -731,7 +733,9 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -865,7 +869,9 @@ def call( ) -> Union[tuple, TFBlipOutput]: # Use BLIP model's config for some fields (if specified) instead of those of vision & text components. output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_outputs = self.vision_model( diff --git a/src/transformers/models/blip/modeling_tf_blip_text.py b/src/transformers/models/blip/modeling_tf_blip_text.py index 7ed4dd013cae..e85d3e627ce9 100644 --- a/src/transformers/models/blip/modeling_tf_blip_text.py +++ b/src/transformers/models/blip/modeling_tf_blip_text.py @@ -853,7 +853,9 @@ def call( `past_key_values`). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if is_decoder: diff --git a/src/transformers/models/bloom/modeling_flax_bloom.py b/src/transformers/models/bloom/modeling_flax_bloom.py index 89f0dab25a52..c7bb1cc9c9a5 100644 --- a/src/transformers/models/bloom/modeling_flax_bloom.py +++ b/src/transformers/models/bloom/modeling_flax_bloom.py @@ -472,7 +472,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/clip/modeling_flax_clip.py b/src/transformers/models/clip/modeling_flax_clip.py index 98102c0369c9..0394974d0647 100644 --- a/src/transformers/models/clip/modeling_flax_clip.py +++ b/src/transformers/models/clip/modeling_flax_clip.py @@ -532,7 +532,9 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(input_ids=input_ids, position_ids=position_ids) @@ -592,7 +594,9 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.embeddings(pixel_values) @@ -671,7 +675,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: @@ -747,7 +753,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -823,7 +831,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if position_ids is None: diff --git a/src/transformers/models/convnext/modeling_tf_convnext.py b/src/transformers/models/convnext/modeling_tf_convnext.py index a91be6221e47..d2e5b0c6f4a6 100644 --- a/src/transformers/models/convnext/modeling_tf_convnext.py +++ b/src/transformers/models/convnext/modeling_tf_convnext.py @@ -369,7 +369,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor]]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -525,7 +527,9 @@ def call( >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ```""" - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -619,7 +623,9 @@ def call( >>> predicted_class_idx = tf.math.argmax(logits, axis=-1)[0] >>> print("Predicted class:", model.config.id2label[int(predicted_class_idx)]) ```""" - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/convnextv2/modeling_tf_convnextv2.py b/src/transformers/models/convnextv2/modeling_tf_convnextv2.py index 70f594c95ae2..9c2484a1ba75 100644 --- a/src/transformers/models/convnextv2/modeling_tf_convnextv2.py +++ b/src/transformers/models/convnextv2/modeling_tf_convnextv2.py @@ -415,7 +415,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor]]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -559,7 +561,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPoolingAndNoAttention, tuple[tf.Tensor]]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -634,7 +638,9 @@ def call( config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py b/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py index fcd7e81c3abd..399da1d9d838 100644 --- a/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py +++ b/src/transformers/models/data2vec/modeling_tf_data2vec_vision.py @@ -789,7 +789,9 @@ def call( training: bool = False, ) -> Union[tuple, TFData2VecVisionModelOutputWithPooling]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1627,7 +1629,9 @@ def call( >>> logits = outputs.logits ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if labels is not None and self.config.num_labels == 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/deit/modeling_tf_deit.py b/src/transformers/models/deit/modeling_tf_deit.py index b45400f8b34e..6dc29ae7fb04 100644 --- a/src/transformers/models/deit/modeling_tf_deit.py +++ b/src/transformers/models/deit/modeling_tf_deit.py @@ -637,7 +637,9 @@ def call( training: bool = False, ) -> Union[TFBaseModelOutputWithPooling, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py b/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py index 7885287d27b5..12d7c942ae50 100644 --- a/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py +++ b/src/transformers/models/deprecated/efficientformer/modeling_tf_efficientformer.py @@ -835,7 +835,9 @@ def call( ) -> Union[TFBaseModelOutput, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/dinov2/modeling_flax_dinov2.py b/src/transformers/models/dinov2/modeling_flax_dinov2.py index 13d564b150d3..b9ea2eaa3ebc 100644 --- a/src/transformers/models/dinov2/modeling_flax_dinov2.py +++ b/src/transformers/models/dinov2/modeling_flax_dinov2.py @@ -600,7 +600,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) diff --git a/src/transformers/models/distilbert/modeling_flax_distilbert.py b/src/transformers/models/distilbert/modeling_flax_distilbert.py index 8c17e2a48b37..fba3dfd9d332 100644 --- a/src/transformers/models/distilbert/modeling_flax_distilbert.py +++ b/src/transformers/models/distilbert/modeling_flax_distilbert.py @@ -467,7 +467,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -508,7 +510,9 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict input_embeds = self.embeddings(input_ids, deterministic=deterministic) diff --git a/src/transformers/models/electra/modeling_flax_electra.py b/src/transformers/models/electra/modeling_flax_electra.py index 23b2e262471a..14d845476d62 100644 --- a/src/transformers/models/electra/modeling_flax_electra.py +++ b/src/transformers/models/electra/modeling_flax_electra.py @@ -786,7 +786,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py b/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py index 2d4bf797b070..13da12b1aa76 100644 --- a/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py +++ b/src/transformers/models/encoder_decoder/modeling_flax_encoder_decoder.py @@ -457,7 +457,9 @@ def encode( >>> encoder_outputs = model.encode(input_ids) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -539,7 +541,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -670,7 +674,9 @@ def __call__( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs diff --git a/src/transformers/models/gemma/modeling_flax_gemma.py b/src/transformers/models/gemma/modeling_flax_gemma.py index bb87a936ddfd..0addcd7dde7a 100644 --- a/src/transformers/models/gemma/modeling_flax_gemma.py +++ b/src/transformers/models/gemma/modeling_flax_gemma.py @@ -494,7 +494,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/gpt2/modeling_flax_gpt2.py b/src/transformers/models/gpt2/modeling_flax_gpt2.py index d22ff5d63f36..8e419217c5a3 100644 --- a/src/transformers/models/gpt2/modeling_flax_gpt2.py +++ b/src/transformers/models/gpt2/modeling_flax_gpt2.py @@ -470,7 +470,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py b/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py index d92eef20a785..a6cdc50b359b 100644 --- a/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py +++ b/src/transformers/models/gpt_neo/modeling_flax_gpt_neo.py @@ -413,7 +413,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/gptj/modeling_flax_gptj.py b/src/transformers/models/gptj/modeling_flax_gptj.py index 42c7c2c4338c..12ea7a4fffb4 100644 --- a/src/transformers/models/gptj/modeling_flax_gptj.py +++ b/src/transformers/models/gptj/modeling_flax_gptj.py @@ -447,7 +447,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/idefics/modeling_tf_idefics.py b/src/transformers/models/idefics/modeling_tf_idefics.py index c6309811ed35..662fe4d88d13 100644 --- a/src/transformers/models/idefics/modeling_tf_idefics.py +++ b/src/transformers/models/idefics/modeling_tf_idefics.py @@ -1282,7 +1282,9 @@ def call( training: Optional[bool] = None, ) -> Union[TFIdeficsBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1710,7 +1712,9 @@ def call( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) diff --git a/src/transformers/models/idefics/vision_tf.py b/src/transformers/models/idefics/vision_tf.py index 67f1208dba9b..9db18f265db5 100644 --- a/src/transformers/models/idefics/vision_tf.py +++ b/src/transformers/models/idefics/vision_tf.py @@ -437,7 +437,7 @@ def call( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) @@ -521,7 +521,7 @@ def call( Returns: """ - + output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions output_hidden_states = ( output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states ) diff --git a/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py b/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py index 87cdeb317fd1..d6703420e4ad 100644 --- a/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py +++ b/src/transformers/models/layoutlmv3/modeling_tf_layoutlmv3.py @@ -979,7 +979,9 @@ def call( # The complexity of this method is mostly just due to handling of these different modalities. output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if input_ids is not None: diff --git a/src/transformers/models/llama/modeling_flax_llama.py b/src/transformers/models/llama/modeling_flax_llama.py index 2b24b5218933..63e34e996ade 100644 --- a/src/transformers/models/llama/modeling_flax_llama.py +++ b/src/transformers/models/llama/modeling_flax_llama.py @@ -476,7 +476,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/longt5/modeling_flax_longt5.py b/src/transformers/models/longt5/modeling_flax_longt5.py index 724a2619bcd6..dee4afeadf72 100644 --- a/src/transformers/models/longt5/modeling_flax_longt5.py +++ b/src/transformers/models/longt5/modeling_flax_longt5.py @@ -1735,7 +1735,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if decoder_input_ids is None: @@ -1833,7 +1835,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1899,7 +1903,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -2285,7 +2291,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/marian/modeling_flax_marian.py b/src/transformers/models/marian/modeling_flax_marian.py index 76aaab32e4c1..e1d9bea4fcdb 100644 --- a/src/transformers/models/marian/modeling_flax_marian.py +++ b/src/transformers/models/marian/modeling_flax_marian.py @@ -990,7 +990,9 @@ def encode( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1061,7 +1063,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1150,7 +1154,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1324,7 +1330,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/marian/modeling_tf_marian.py b/src/transformers/models/marian/modeling_tf_marian.py index e0aac8c3e5a3..9b03424444f2 100644 --- a/src/transformers/models/marian/modeling_tf_marian.py +++ b/src/transformers/models/marian/modeling_tf_marian.py @@ -1168,7 +1168,9 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/mbart/modeling_flax_mbart.py b/src/transformers/models/mbart/modeling_flax_mbart.py index dc0ac48b4efe..ad269860a959 100644 --- a/src/transformers/models/mbart/modeling_flax_mbart.py +++ b/src/transformers/models/mbart/modeling_flax_mbart.py @@ -1064,7 +1064,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1134,7 +1136,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1223,7 +1227,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1395,7 +1401,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/mbart/modeling_tf_mbart.py b/src/transformers/models/mbart/modeling_tf_mbart.py index 3f64655cb0fa..44ac6562a861 100644 --- a/src/transformers/models/mbart/modeling_tf_mbart.py +++ b/src/transformers/models/mbart/modeling_tf_mbart.py @@ -1183,7 +1183,9 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if decoder_input_ids is None and input_ids is not None: decoder_input_ids = shift_tokens_right(input_ids, self.config.pad_token_id) diff --git a/src/transformers/models/mistral/modeling_flax_mistral.py b/src/transformers/models/mistral/modeling_flax_mistral.py index e36d1536fcde..5b34a5144168 100644 --- a/src/transformers/models/mistral/modeling_flax_mistral.py +++ b/src/transformers/models/mistral/modeling_flax_mistral.py @@ -470,7 +470,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict batch_size, sequence_length = input_ids.shape diff --git a/src/transformers/models/mobilevit/modeling_tf_mobilevit.py b/src/transformers/models/mobilevit/modeling_tf_mobilevit.py index f7fc9ba30123..4acf1ff013ac 100644 --- a/src/transformers/models/mobilevit/modeling_tf_mobilevit.py +++ b/src/transformers/models/mobilevit/modeling_tf_mobilevit.py @@ -834,7 +834,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPooling]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # When running on CPU, `keras.layers.Conv2D` doesn't support `NCHW` format. @@ -1316,7 +1318,9 @@ def call( >>> # logits are of shape (batch_size, num_labels, height, width) >>> logits = outputs.logits ```""" - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None and not self.config.num_labels > 1: diff --git a/src/transformers/models/opt/modeling_flax_opt.py b/src/transformers/models/opt/modeling_flax_opt.py index e22af5e1c4b2..d2f77ecbee26 100644 --- a/src/transformers/models/opt/modeling_flax_opt.py +++ b/src/transformers/models/opt/modeling_flax_opt.py @@ -594,7 +594,9 @@ def __call__( deterministic: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: diff --git a/src/transformers/models/opt/modeling_tf_opt.py b/src/transformers/models/opt/modeling_tf_opt.py index 930064298503..f403d1cc5cc4 100644 --- a/src/transformers/models/opt/modeling_tf_opt.py +++ b/src/transformers/models/opt/modeling_tf_opt.py @@ -634,7 +634,9 @@ def call( behaviors between training and evaluation). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -788,7 +790,9 @@ def call( **kwargs, ) -> Union[TFBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -866,7 +870,9 @@ def call( **kwargs, ) -> Union[TFBaseModelOutputWithPast, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1027,7 +1033,9 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.model( diff --git a/src/transformers/models/pegasus/modeling_flax_pegasus.py b/src/transformers/models/pegasus/modeling_flax_pegasus.py index 734b084d8086..ddf0ae492407 100644 --- a/src/transformers/models/pegasus/modeling_flax_pegasus.py +++ b/src/transformers/models/pegasus/modeling_flax_pegasus.py @@ -1007,7 +1007,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1078,7 +1080,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1167,7 +1171,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1342,7 +1348,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/pegasus/modeling_tf_pegasus.py b/src/transformers/models/pegasus/modeling_tf_pegasus.py index ec11449b086e..48f9e0b6953f 100644 --- a/src/transformers/models/pegasus/modeling_tf_pegasus.py +++ b/src/transformers/models/pegasus/modeling_tf_pegasus.py @@ -1183,7 +1183,9 @@ def call( if decoder_input_ids is None and decoder_inputs_embeds is None: use_cache = False - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if encoder_outputs is None: encoder_outputs = self.encoder( diff --git a/src/transformers/models/regnet/modeling_flax_regnet.py b/src/transformers/models/regnet/modeling_flax_regnet.py index 69882dd43651..2cc3707fa51a 100644 --- a/src/transformers/models/regnet/modeling_flax_regnet.py +++ b/src/transformers/models/regnet/modeling_flax_regnet.py @@ -603,7 +603,9 @@ def __call__( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ): - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -647,7 +649,9 @@ def __call__( output_hidden_states: bool = False, return_dict: bool = True, ) -> FlaxBaseModelOutputWithPoolingAndNoAttention: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, deterministic=deterministic) diff --git a/src/transformers/models/regnet/modeling_tf_regnet.py b/src/transformers/models/regnet/modeling_tf_regnet.py index 81d153829813..b06e64b3b406 100644 --- a/src/transformers/models/regnet/modeling_tf_regnet.py +++ b/src/transformers/models/regnet/modeling_tf_regnet.py @@ -395,7 +395,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> TFBaseModelOutputWithPoolingAndNoAttention: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, training=training) @@ -504,7 +506,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[TFBaseModelOutputWithPoolingAndNoAttention, tuple[tf.Tensor]]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.regnet( @@ -570,7 +574,9 @@ def call( Labels for computing the image classification/regression loss. Indices should be in `[0, ..., config.num_labels - 1]`. If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.regnet( diff --git a/src/transformers/models/resnet/modeling_flax_resnet.py b/src/transformers/models/resnet/modeling_flax_resnet.py index ac4a3e94f5d9..a2a9418b7cf2 100644 --- a/src/transformers/models/resnet/modeling_flax_resnet.py +++ b/src/transformers/models/resnet/modeling_flax_resnet.py @@ -494,7 +494,9 @@ def __call__( output_hidden_states: Optional[bool] = None, return_dict: Optional[bool] = None, ): - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict pixel_values = jnp.transpose(pixel_values, (0, 2, 3, 1)) @@ -537,7 +539,9 @@ def __call__( output_hidden_states: bool = False, return_dict: bool = True, ) -> FlaxBaseModelOutputWithPoolingAndNoAttention: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict embedding_output = self.embedder(pixel_values, deterministic=deterministic) diff --git a/src/transformers/models/resnet/modeling_tf_resnet.py b/src/transformers/models/resnet/modeling_tf_resnet.py index 2ef260710481..f7c415f97b05 100644 --- a/src/transformers/models/resnet/modeling_tf_resnet.py +++ b/src/transformers/models/resnet/modeling_tf_resnet.py @@ -420,7 +420,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPoolingAndNoAttention]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 image layers can't use NCHW format when running on CPU. @@ -493,7 +495,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple[tf.Tensor], TFBaseModelOutputWithPoolingAndNoAttention]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict resnet_outputs = self.resnet( diff --git a/src/transformers/models/roberta/modeling_flax_roberta.py b/src/transformers/models/roberta/modeling_flax_roberta.py index fd4383c68de5..3b46c0fa682f 100644 --- a/src/transformers/models/roberta/modeling_flax_roberta.py +++ b/src/transformers/models/roberta/modeling_flax_roberta.py @@ -833,7 +833,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py b/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py index add14852e368..f65dc07bb165 100644 --- a/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py +++ b/src/transformers/models/roberta_prelayernorm/modeling_flax_roberta_prelayernorm.py @@ -837,7 +837,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/roformer/modeling_flax_roformer.py b/src/transformers/models/roformer/modeling_flax_roformer.py index 57327c97ba1f..de78eb4787c0 100644 --- a/src/transformers/models/roformer/modeling_flax_roformer.py +++ b/src/transformers/models/roformer/modeling_flax_roformer.py @@ -656,7 +656,9 @@ def __call__( return_dict: Optional[bool] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed diff --git a/src/transformers/models/sam/modeling_tf_sam.py b/src/transformers/models/sam/modeling_tf_sam.py index d9f071501257..c44dfdf80ac8 100644 --- a/src/transformers/models/sam/modeling_tf_sam.py +++ b/src/transformers/models/sam/modeling_tf_sam.py @@ -436,7 +436,9 @@ def call( return_dict: Optional[bool] = None, ) -> Union[tuple, TFBaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_attentions = () @@ -1272,7 +1274,9 @@ def call( training: Optional[bool] = False, ) -> Union[tuple, TFSamVisionEncoderOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1579,7 +1583,9 @@ def call( **kwargs, ) -> TFSamImageSegmentationOutput | tuple[tf.Tensor]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None and image_embeddings is None: diff --git a/src/transformers/models/segformer/modeling_tf_segformer.py b/src/transformers/models/segformer/modeling_tf_segformer.py index 5f759f8de217..e3301b4dfc6a 100644 --- a/src/transformers/models/segformer/modeling_tf_segformer.py +++ b/src/transformers/models/segformer/modeling_tf_segformer.py @@ -586,7 +586,9 @@ def call( training: bool = False, ) -> Union[tuple, TFBaseModelOutput]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # When running on CPU, `keras.layers.Conv2D` doesn't support `NCHW` format. @@ -982,7 +984,9 @@ def call( [1, 150, 128, 128] ```""" return_dict = return_dict if return_dict is not None else self.config.use_return_dict - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) if labels is not None and not self.config.num_labels > 1: raise ValueError("The number of labels should be greater than one") diff --git a/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py b/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py index 0a42cfc8218f..d2d23d29007f 100644 --- a/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py +++ b/src/transformers/models/speech_encoder_decoder/modeling_flax_speech_encoder_decoder.py @@ -494,7 +494,9 @@ def encode( >>> encoder_outputs = model.encode(inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -572,7 +574,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -696,7 +700,9 @@ def __call__( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs diff --git a/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py b/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py index c80a8ba807fe..fad6a026b152 100755 --- a/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py +++ b/src/transformers/models/speech_to_text/modeling_tf_speech_to_text.py @@ -1235,7 +1235,9 @@ def call( **kwargs, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/swiftformer/modeling_tf_swiftformer.py b/src/transformers/models/swiftformer/modeling_tf_swiftformer.py index fd08d5df870f..612c2406a1d0 100644 --- a/src/transformers/models/swiftformer/modeling_tf_swiftformer.py +++ b/src/transformers/models/swiftformer/modeling_tf_swiftformer.py @@ -550,7 +550,9 @@ def call( return_dict: Optional[bool] = None, training: bool = False, ) -> Union[tuple, TFBaseModelOutputWithNoAttention]: - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict all_hidden_states = (hidden_states,) if output_hidden_states else None @@ -657,7 +659,9 @@ def call( ) -> Union[tuple, TFBaseModelOutputWithNoAttention]: r""" """ - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 image layers can't use NCHW format when running on CPU. diff --git a/src/transformers/models/swin/modeling_tf_swin.py b/src/transformers/models/swin/modeling_tf_swin.py index 276a1ff0d19d..5b9f9f272ff4 100644 --- a/src/transformers/models/swin/modeling_tf_swin.py +++ b/src/transformers/models/swin/modeling_tf_swin.py @@ -1242,7 +1242,9 @@ def call( training: bool = False, ) -> Union[TFSwinModelOutput, tuple[tf.Tensor, ...]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: @@ -1340,7 +1342,9 @@ def call( Boolean masked positions. Indicates which patches are masked (1) and which aren't (0). """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if pixel_values is None: diff --git a/src/transformers/models/t5/modeling_flax_t5.py b/src/transformers/models/t5/modeling_flax_t5.py index 734d77924f90..c829a084e4d9 100644 --- a/src/transformers/models/t5/modeling_flax_t5.py +++ b/src/transformers/models/t5/modeling_flax_t5.py @@ -997,7 +997,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if decoder_input_ids is None: @@ -1095,7 +1097,9 @@ def encode( >>> encoder_outputs = model.encode(**inputs) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if attention_mask is None: @@ -1161,7 +1165,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1460,7 +1466,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare encoder inputs @@ -1635,7 +1643,9 @@ def decode( >>> logits = outputs.logits ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] diff --git a/src/transformers/models/whisper/modeling_flax_whisper.py b/src/transformers/models/whisper/modeling_flax_whisper.py index bcc25bdb9443..8cb98f2385d0 100644 --- a/src/transformers/models/whisper/modeling_flax_whisper.py +++ b/src/transformers/models/whisper/modeling_flax_whisper.py @@ -991,7 +991,9 @@ def encode( >>> encoder_outputs = model.encode(input_features=input_features) ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # Handle any PRNG if needed @@ -1056,7 +1058,9 @@ def decode( ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1144,7 +1148,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # prepare decoder inputs @@ -1303,7 +1309,9 @@ def decode( >>> last_decoder_hidden_states = outputs.last_hidden_state ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict encoder_hidden_states = encoder_outputs[0] @@ -1551,7 +1559,9 @@ def __call__( return_dict: bool = True, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict if encoder_outputs is None: @@ -1626,7 +1636,9 @@ def __call__( **kwargs, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # Handle any PRNG if needed diff --git a/src/transformers/models/whisper/modeling_tf_whisper.py b/src/transformers/models/whisper/modeling_tf_whisper.py index caf1eca76260..f5ca846b81e3 100644 --- a/src/transformers/models/whisper/modeling_tf_whisper.py +++ b/src/transformers/models/whisper/modeling_tf_whisper.py @@ -749,7 +749,9 @@ def call( """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.use_return_dict # TF 2.0 layers can't use channels first format when running on CPU. @@ -966,7 +968,9 @@ def call( Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1152,7 +1156,9 @@ def call( [1, 2, 512] ```""" output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/xglm/modeling_flax_xglm.py b/src/transformers/models/xglm/modeling_flax_xglm.py index 44d7c273ac94..1366148d9a3d 100644 --- a/src/transformers/models/xglm/modeling_flax_xglm.py +++ b/src/transformers/models/xglm/modeling_flax_xglm.py @@ -624,7 +624,9 @@ def __call__( dropout_rng: PRNGKey = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict if encoder_hidden_states is not None and encoder_attention_mask is None: diff --git a/src/transformers/models/xglm/modeling_tf_xglm.py b/src/transformers/models/xglm/modeling_tf_xglm.py index e7c6614f72e7..4ca60b5e5ad6 100644 --- a/src/transformers/models/xglm/modeling_tf_xglm.py +++ b/src/transformers/models/xglm/modeling_tf_xglm.py @@ -535,7 +535,9 @@ def call( **kwargs: Any, ) -> Union[TFBaseModelOutputWithPastAndCrossAttentions, tuple[tf.Tensor]]: output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict diff --git a/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py b/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py index 3f0d34e5c9b7..bdbc06620a1b 100644 --- a/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py +++ b/src/transformers/models/xlm_roberta/modeling_flax_xlm_roberta.py @@ -840,7 +840,9 @@ def __call__( past_key_values: Optional[dict] = None, ): output_attentions = output_attentions if output_attentions is not None else self.config.output_attentions - + output_hidden_states = ( + output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states + ) return_dict = return_dict if return_dict is not None else self.config.return_dict # init input tensors if not passed From a4ec22ae4d8473ea2bd63e45b95340161200c98a Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 14:25:44 +0800 Subject: [PATCH 43/51] Most probably explicit register of the attention classes to prevent holding the references in the decoder layer. Will do that in a bit --- src/transformers/modeling_utils.py | 10 +---- .../modeling_audio_spectrogram_transformer.py | 40 +++++++------------ .../models/llama/modeling_llama.py | 7 +--- src/transformers/utils/generic.py | 22 ++++------ 4 files changed, 26 insertions(+), 53 deletions(-) diff --git a/src/transformers/modeling_utils.py b/src/transformers/modeling_utils.py index 285a2fe0c70c..bfd283d4eddf 100644 --- a/src/transformers/modeling_utils.py +++ b/src/transformers/modeling_utils.py @@ -33,7 +33,7 @@ from enum import Enum from functools import partial, wraps from threading import Thread -from typing import Any, Callable, Optional, TypeVar, Union +from typing import Any, Callable, Dict, Optional, Tuple, TypeVar, Union from zipfile import is_zipfile import torch @@ -75,13 +75,7 @@ from .loss.loss_utils import LOSS_MAPPING from .modeling_layers import GradientCheckpointingLayer from .pytorch_utils import ( # noqa: F401 - Conv1D, - apply_chunking_to_forward, - find_pruneable_heads_and_indices, id_tensor_storage, - prune_conv1d_layer, - prune_layer, - prune_linear_layer, ) from .quantizers import AutoHfQuantizer, HfQuantizer from .quantizers.quantizers_utils import get_module_from_name @@ -2052,7 +2046,7 @@ def __init__(self, config: PretrainedConfig, *inputs, **kwargs): self._keep_in_fp32_modules = copy.copy(self.__class__._keep_in_fp32_modules) self._no_split_modules = self._no_split_modules or [] - self._can_record_outputs = { + self._can_record_outputs: Dict[str, Tuple[nn.Module, int]] = { "hidden_states": (GradientCheckpointingLayer, 0), "attentions": (GradientCheckpointingLayer, 1), } diff --git a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py index 8ab567fa8320..67a8356dd7c3 100644 --- a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py +++ b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py @@ -148,45 +148,36 @@ def __init__(self, config: ASTConfig) -> None: self.key = nn.Linear(config.hidden_size, self.all_head_size, bias=config.qkv_bias) self.value = nn.Linear(config.hidden_size, self.all_head_size, bias=config.qkv_bias) - def transpose_for_scores(self, x: torch.Tensor) -> torch.Tensor: - new_x_shape = x.size()[:-1] + (self.num_attention_heads, self.attention_head_size) - x = x.view(new_x_shape) - return x.permute(0, 2, 1, 3) - def forward( - self, hidden_states, head_mask: Optional[torch.Tensor] = None, output_attentions: bool = False + self, hidden_states, head_mask: Optional[torch.Tensor] = None, **kwargs ) -> Union[tuple[torch.Tensor, torch.Tensor], tuple[torch.Tensor]]: - key_layer = self.transpose_for_scores(self.key(hidden_states)) - value_layer = self.transpose_for_scores(self.value(hidden_states)) - query_layer = self.transpose_for_scores(self.query(hidden_states)) + input_shape = hidden_states.shape[:-1] + hidden_shape = (*input_shape, -1, self.head_dim) + + query_states = self.query(hidden_states).view(hidden_shape).transpose(1, 2) + key_states = self.key(hidden_states).view(hidden_shape).transpose(1, 2) + value_states = self.value(hidden_states).view(hidden_shape).transpose(1, 2) attention_interface: Callable = eager_attention_forward if self.config._attn_implementation != "eager": - if self.config._attn_implementation == "sdpa" and output_attentions: - logger.warning_once( - "`torch.nn.functional.scaled_dot_product_attention` does not support `output_attentions=True`. Falling back to " - 'eager attention. This warning can be removed using the argument `attn_implementation="eager"` when loading the model.' - ) - else: - attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] + attention_interface = ALL_ATTENTION_FUNCTIONS[self.config._attn_implementation] context_layer, attention_probs = attention_interface( self, - query_layer, - key_layer, - value_layer, + query_states, + key_states, + value_states, head_mask, is_causal=self.is_causal, scaling=self.scaling, dropout=0.0 if not self.training else self.dropout_prob, + *kwargs, ) new_context_layer_shape = context_layer.size()[:-2] + (self.all_head_size,) context_layer = context_layer.reshape(new_context_layer_shape) - outputs = (context_layer, attention_probs) if output_attentions else (context_layer,) - - return outputs + return context_layer, attention_probs # Copied from transformers.models.vit.modeling_vit.ViTSelfOutput with ViT->AST @@ -436,7 +427,6 @@ def forward( self, input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" @@ -447,8 +437,7 @@ def forward( [`AutoFeatureExtractor`] should be used for extracting the mel features, padding and conversion into a tensor of type `torch.FloatTensor`. See [`~ASTFeatureExtractor.__call__`] """ - - + return_dict = return_dict if return_dict is not None else self.config.use_return_dict if input_values is None: @@ -523,7 +512,6 @@ def forward( input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, SequenceClassifierOutput]: r""" diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index c4bf565e78b6..ac6b4790f6d2 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -229,7 +229,7 @@ def forward( past_key_value: Optional[Cache] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], - ) -> tuple[torch.Tensor, Optional[torch.Tensor], Optional[tuple[torch.Tensor]]]: + ) -> tuple[torch.Tensor, torch.Tensor]: input_shape = hidden_states.shape[:-1] hidden_shape = (*input_shape, -1, self.head_dim) @@ -287,7 +287,7 @@ def forward( cache_position: Optional[torch.LongTensor] = None, position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None, # necessary, but kept here for BC **kwargs: Unpack[FlashAttentionKwargs], - ) -> tuple[torch.FloatTensor, Optional[tuple[torch.FloatTensor, torch.FloatTensor]]]: + ) -> tuple[torch.FloatTensor, torch.Tensor]: residual = hidden_states hidden_states = self.input_layernorm(hidden_states) @@ -571,7 +571,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, ) -> SequenceClassifierOutputWithPast: r""" @@ -660,7 +659,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, start_positions: Optional[torch.LongTensor] = None, end_positions: Optional[torch.LongTensor] = None, - **kwargs, ) -> QuestionAnsweringModelOutput: outputs: BaseModelOutputWithPast = self.transformer( @@ -728,7 +726,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, ) -> TokenClassifierOutput: r""" diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index b83ae118b855..7381b1e44928 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -966,8 +966,6 @@ def wrapper(self, *args, **kwargs): return_dict = kwargs.pop("return_dict", self.config.use_return_dict) # Safely extract common config-backed flags - kwargs.setdefault("output_attentions", output_attentions) - kwargs.setdefault("output_hidden_states", output_hidden_states) kwargs.setdefault("use_cache", use_cache) kwargs["return_dict"] = kwargs.pop("return_dict", return_dict) @@ -980,26 +978,22 @@ def wrapper(self, *args, **kwargs): # Extract input_ids and inputs_embeds more robustly input_ids = all_args.get("input_ids", None) inputs_embeds = all_args.get("inputs_embeds", None) + # TODO check the inputs ids vs input embeds. + # probably a key string saying which input are not compatible with each other use_cache = kwargs["use_cache"] past_key_values = kwargs.get("past_key_values", None) hooks = [] - collected_attentions = [] - collected_hidden_states = [] - if output_attentions or output_hidden_states: - - def output_hidden_and_attention(module, inp, out): - if output_hidden_states: - collected_hidden_states.append(out[0]) - if output_attentions: - collected_attentions.append(out[1]) + collected_outputs = {} def make_capture_fn(key, index): def capture_fn(module, input, output): - collected[key].append(output[index]) + collected_outputs[key].append(output[index]) return capture_fn + capture_flags = self._can_record_outputs.keys() + recordable_keys = {k.replace("output_"): v for k, v in capture_flags.items()} if any(capture_flags.values()): for _, layer in self.named_modules(): for key, (cls, idx) in self._can_record_outputs.items(): @@ -1013,10 +1007,10 @@ def capture_fn(module, input, output): h.remove() # Insert collected outputs - for key in collected: + for key in collected_outputs: # Optional: append a final model-level output like `last_hidden_state` for hidden_states # if needed. You can customize this as per key semantics. - outputs[key] = tuple(collected[key]) if collected[key] else None + outputs[key] = tuple(collected_outputs[key]) if collected_outputs[key] else None # Fill missing keys with None if not using return_dict if return_dict is False: From f06e39ed698151f2c1605aac4511fefabbdb7dff Mon Sep 17 00:00:00 2001 From: Arthur Date: Sun, 22 Jun 2025 16:02:59 +0800 Subject: [PATCH 44/51] some more cleanup --- .../modeling_audio_spectrogram_transformer.py | 102 +--- .../models/autoformer/modeling_autoformer.py | 14 +- .../models/aya_vision/modular_aya_vision.py | 18 - .../models/bamba/modular_bamba.py | 65 +-- src/transformers/models/bart/modeling_bart.py | 458 +----------------- src/transformers/utils/generic.py | 4 - 6 files changed, 46 insertions(+), 615 deletions(-) diff --git a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py index 67a8356dd7c3..e81f6e2c7745 100644 --- a/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py +++ b/src/transformers/models/audio_spectrogram_transformer/modeling_audio_spectrogram_transformer.py @@ -17,9 +17,10 @@ from typing import Callable, Optional, Union import torch -import torch.utils.checkpoint from torch import nn -from torch.nn import BCEWithLogitsLoss, CrossEntropyLoss, MSELoss + +from transformers.modeling_layers import GradientCheckpointingLayer +from transformers.utils.generic import check_model_inputs from ...activations import ACT2FN from ...modeling_outputs import BaseModelOutput, BaseModelOutputWithPooling, SequenceClassifierOutput @@ -192,7 +193,7 @@ def __init__(self, config: ASTConfig) -> None: self.dense = nn.Linear(config.hidden_size, config.hidden_size) self.dropout = nn.Dropout(config.hidden_dropout_prob) - def forward(self, hidden_states: torch.Tensor, input_tensor: torch.Tensor) -> torch.Tensor: + def forward(self, hidden_states: torch.Tensor) -> torch.Tensor: hidden_states = self.dense(hidden_states) hidden_states = self.dropout(hidden_states) @@ -229,14 +230,10 @@ def forward( self, hidden_states: torch.Tensor, head_mask: Optional[torch.Tensor] = None, - output_attentions: bool = False, ) -> Union[tuple[torch.Tensor, torch.Tensor], tuple[torch.Tensor]]: self_outputs = self.attention(hidden_states, head_mask, output_attentions) - attention_output = self.output(self_outputs[0], hidden_states) - - outputs = (attention_output,) + self_outputs[1:] # add attentions if we output them - return outputs + return attention_output # Copied from transformers.models.vit.modeling_vit.ViTIntermediate with ViT->AST @@ -273,7 +270,7 @@ def forward(self, hidden_states: torch.Tensor, input_tensor: torch.Tensor) -> to # Copied from transformers.models.vit.modeling_vit.ViTLayer with ViT->AST,VIT->AST -class ASTLayer(nn.Module): +class ASTLayer(GradientCheckpointingLayer): """This corresponds to the Block class in the timm implementation.""" def __init__(self, config: ASTConfig) -> None: @@ -291,14 +288,13 @@ def forward( hidden_states: torch.Tensor, head_mask: Optional[torch.Tensor] = None, output_attentions: bool = False, - ) -> Union[tuple[torch.Tensor, torch.Tensor], tuple[torch.Tensor]]: + ) -> torch.Tensor: self_attention_outputs = self.attention( self.layernorm_before(hidden_states), # in AST, layernorm is applied before self-attention head_mask, output_attentions=output_attentions, ) attention_output = self_attention_outputs[0] - outputs = self_attention_outputs[1:] # add self attentions if we output attention weights # first residual connection hidden_states = attention_output + hidden_states @@ -309,10 +305,7 @@ def forward( # second residual connection is done here layer_output = self.output(layer_output, hidden_states) - - outputs = (layer_output,) + outputs - - return outputs + return layer_output # Copied from transformers.models.vit.modeling_vit.ViTEncoder with ViT->AST @@ -323,47 +316,18 @@ def __init__(self, config: ASTConfig) -> None: self.layer = nn.ModuleList([ASTLayer(config) for _ in range(config.num_hidden_layers)]) self.gradient_checkpointing = False + @check_model_inputs def forward( self, hidden_states: torch.Tensor, head_mask: Optional[torch.Tensor] = None, - output_attentions: bool = False, - output_hidden_states: bool = False, - return_dict: bool = True, - ) -> Union[tuple, BaseModelOutput]: - all_hidden_states = () if output_hidden_states else None - all_self_attentions = () if output_attentions else None - + ) -> BaseModelOutput: for i, layer_module in enumerate(self.layer): - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - layer_head_mask = head_mask[i] if head_mask is not None else None - - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - layer_module.__call__, - hidden_states, - layer_head_mask, - output_attentions, - ) - else: - layer_outputs = layer_module(hidden_states, layer_head_mask, output_attentions) - + layer_outputs = layer_module(hidden_states, layer_head_mask) hidden_states = layer_outputs[0] - - if output_attentions: - all_self_attentions = all_self_attentions + (layer_outputs[1],) - - if output_hidden_states: - all_hidden_states = all_hidden_states + (hidden_states,) - - if not return_dict: - return tuple(v for v in [hidden_states, all_hidden_states, all_self_attentions] if v is not None) return BaseModelOutput( last_hidden_state=hidden_states, - hidden_states=all_hidden_states, - attentions=all_self_attentions, ) @@ -427,7 +391,6 @@ def forward( self, input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutputWithPooling]: r""" input_values (`torch.FloatTensor` of shape `(batch_size, max_length, num_mel_bins)`): @@ -438,8 +401,6 @@ def forward( tensor of type `torch.FloatTensor`. See [`~ASTFeatureExtractor.__call__`] """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if input_values is None: raise ValueError("You have to specify input_values") @@ -455,23 +416,15 @@ def forward( encoder_outputs = self.encoder( embedding_output, head_mask=head_mask, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) sequence_output = encoder_outputs[0] sequence_output = self.layernorm(sequence_output) pooled_output = (sequence_output[:, 0] + sequence_output[:, 1]) / 2 - if not return_dict: - return (sequence_output, pooled_output) + encoder_outputs[1:] - return BaseModelOutputWithPooling( last_hidden_state=sequence_output, pooler_output=pooled_output, - hidden_states=encoder_outputs.hidden_states, - attentions=encoder_outputs.attentions, ) @@ -512,7 +465,7 @@ def forward( input_values: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, labels: Optional[torch.Tensor] = None, - return_dict: Optional[bool] = None, + **kwargs, ) -> Union[tuple, SequenceClassifierOutput]: r""" input_values (`torch.FloatTensor` of shape `(batch_size, max_length, num_mel_bins)`): @@ -526,14 +479,10 @@ def forward( config.num_labels - 1]`. If `config.num_labels == 1` a regression loss is computed (Mean-Square loss), If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict outputs = self.audio_spectrogram_transformer( input_values, head_mask=head_mask, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) pooled_output = outputs[1] @@ -541,36 +490,11 @@ def forward( loss = None if labels is not None: - if self.config.problem_type is None: - if self.num_labels == 1: - self.config.problem_type = "regression" - elif self.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int): - self.config.problem_type = "single_label_classification" - else: - self.config.problem_type = "multi_label_classification" - - if self.config.problem_type == "regression": - loss_fct = MSELoss() - if self.num_labels == 1: - loss = loss_fct(logits.squeeze(), labels.squeeze()) - else: - loss = loss_fct(logits, labels) - elif self.config.problem_type == "single_label_classification": - loss_fct = CrossEntropyLoss() - loss = loss_fct(logits.view(-1, self.num_labels), labels.view(-1)) - elif self.config.problem_type == "multi_label_classification": - loss_fct = BCEWithLogitsLoss() - loss = loss_fct(logits, labels) - - if not return_dict: - output = (logits,) + outputs[2:] - return ((loss,) + output) if loss is not None else output + self.loss_func(input_values, labels, **kwargs) return SequenceClassifierOutput( loss=loss, logits=logits, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) diff --git a/src/transformers/models/autoformer/modeling_autoformer.py b/src/transformers/models/autoformer/modeling_autoformer.py index c8d5c5221897..58c3e6e6ccdb 100644 --- a/src/transformers/models/autoformer/modeling_autoformer.py +++ b/src/transformers/models/autoformer/modeling_autoformer.py @@ -16,6 +16,8 @@ # limitations under the License. """PyTorch Autoformer model.""" +# TODO to refactor the attention is absolutely shit in this modeling code. + import math from dataclasses import dataclass from typing import Optional, Union @@ -969,7 +971,6 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" @@ -1000,8 +1001,7 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - - + return_dict = return_dict if return_dict is not None else self.config.use_return_dict hidden_states = self.value_embedding(inputs_embeds) @@ -1110,7 +1110,6 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, ) -> Union[tuple, AutoFormerDecoderOutput]: r""" @@ -1175,8 +1174,7 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - - + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1583,8 +1581,7 @@ def forward( >>> last_hidden_state = outputs.last_hidden_state ```""" - - + use_cache = use_cache if use_cache is not None else self.config.use_cache return_dict = return_dict if return_dict is not None else self.config.use_return_dict @@ -1985,7 +1982,6 @@ def generate( past_observed_mask: Optional[torch.Tensor] = None, static_categorical_features: Optional[torch.Tensor] = None, static_real_features: Optional[torch.Tensor] = None, - ) -> SampleTSPredictionOutput: r""" Greedily generate sequences of sample predictions from a model with a probability distribution head. diff --git a/src/transformers/models/aya_vision/modular_aya_vision.py b/src/transformers/models/aya_vision/modular_aya_vision.py index bc54cc351817..38628adb26a6 100644 --- a/src/transformers/models/aya_vision/modular_aya_vision.py +++ b/src/transformers/models/aya_vision/modular_aya_vision.py @@ -186,17 +186,9 @@ def forward( vision_feature_layer: Optional[Union[int, list[int]]] = None, vision_feature_select_strategy: Optional[str] = None, use_cache: Optional[bool] = None, - output_attentions: Optional[bool] = None, - output_hidden_states: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AyaVisionModelOutputWithPast]: - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer ) @@ -244,9 +236,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=True, cache_position=cache_position, **kwargs, ) @@ -254,8 +243,6 @@ def forward( return AyaVisionModelOutputWithPast( last_hidden_state=outputs.last_hidden_state, past_key_values=outputs.past_key_values, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, image_hidden_states=image_features if pixel_values is not None else None, ) @@ -273,8 +260,6 @@ def forward( vision_feature_select_strategy: Optional[str] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, image_sizes: Optional[torch.Tensor] = None, @@ -327,9 +312,6 @@ def forward( vision_feature_select_strategy=vision_feature_select_strategy, labels=labels, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, logits_to_keep=logits_to_keep, image_sizes=image_sizes, diff --git a/src/transformers/models/bamba/modular_bamba.py b/src/transformers/models/bamba/modular_bamba.py index 83b9655f5b71..cb1517a21a19 100644 --- a/src/transformers/models/bamba/modular_bamba.py +++ b/src/transformers/models/bamba/modular_bamba.py @@ -22,7 +22,6 @@ from typing import Optional, TypedDict, Union import torch -import torch.utils.checkpoint from torch import nn import transformers.models.jamba.modeling_jamba as modeling_jamba @@ -49,7 +48,7 @@ from ...processing_utils import Unpack from ...utils import ( auto_docstring, - can_return_tuple, + check_model_inputs, logging, ) from ...utils.import_utils import is_causal_conv1d_available, is_flash_attn_2_available, is_mamba_2_ssm_available @@ -735,7 +734,6 @@ def forward( attention_mask: Optional[torch.Tensor] = None, position_ids: Optional[torch.LongTensor] = None, past_key_value: Optional[HybridMambaAttentionDynamicCache] = None, - output_attentions: Optional[bool] = False, use_cache: Optional[bool] = False, cache_position: Optional[torch.LongTensor] = None, position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None, # necessary, but kept here for BC @@ -778,17 +776,16 @@ def forward( ) self_attn_weights = None elif self.layer_type == "attention": - hidden_states, self_attn_weights = self.self_attn( + hidden_states = self.self_attn( hidden_states=hidden_states, attention_mask=attention_mask, position_ids=position_ids, past_key_value=past_key_value, - output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, **kwargs, - ) + )[0] # residual connection after attention hidden_states = residual + hidden_states @@ -799,12 +796,7 @@ def forward( hidden_states = self.feed_forward(hidden_states) hidden_states = residual + hidden_states - outputs = (hidden_states,) - - if output_attentions: - outputs += (self_attn_weights,) - - return outputs + return hidden_states @auto_docstring @@ -864,7 +856,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value - @can_return_tuple + @check_model_inputs @auto_docstring def forward( self, @@ -874,35 +866,16 @@ def forward( past_key_values: Optional[HybridMambaAttentionDynamicCache] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[BambaFlashAttentionKwargs], ) -> BaseModelOutputWithPast: - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) - use_cache = use_cache if use_cache is not None else self.config.use_cache - if (input_ids is None) ^ (inputs_embeds is not None): raise ValueError("You must specify exactly one of input_ids or inputs_embeds") - if self.gradient_checkpointing and self.training and use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`." - ) - use_cache = False - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) hidden_states = inputs_embeds - if use_cache and past_key_values is None: - logger.warning_once( - "Bamba requires an initialized `HybridMambaAttentionDynamicCache` to return a cache. None was " - "provided, so no cache will be returned." - ) - if cache_position is None: cache_position = torch.arange(hidden_states.shape[1], device=hidden_states.device) if position_ids is None: @@ -916,22 +889,15 @@ def forward( # create position embeddings to be shared across the decoder layers position_embeddings = self.rotary_emb(hidden_states, position_ids) - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - for decoder_layer in self.layers: # Depending on the layer type we opt for 2D base attention mask (Mamba) or 4D causal mask (Attention) layer_mask = mamba_mask if decoder_layer.layer_type == "mamba" else causal_mask - if output_hidden_states: - all_hidden_states += (hidden_states,) - layer_outputs = decoder_layer( hidden_states, attention_mask=layer_mask, position_ids=position_ids, past_key_value=past_key_values, - output_attentions=output_attentions, use_cache=use_cache, cache_position=cache_position, position_embeddings=position_embeddings, @@ -940,17 +906,7 @@ def forward( hidden_states = layer_outputs[0] - if output_attentions: - if layer_outputs[1] is not None: - # append attentions only of attention layers. Mamba layers return `None` as the attention weights - all_self_attns += (layer_outputs[1],) - hidden_states = self.final_layernorm(hidden_states) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - if past_key_values and not past_key_values.has_previous_state: past_key_values.has_previous_state = True @@ -959,8 +915,6 @@ def forward( return BaseModelOutputWithPast( last_hidden_state=hidden_states, past_key_values=next_cache, - hidden_states=all_hidden_states, - attentions=all_self_attns, ) def _update_causal_mask( @@ -1108,7 +1062,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - cache_position: Optional[torch.LongTensor] = None, logits_to_keep: Union[int, torch.Tensor] = 0, **kwargs, @@ -1135,10 +1088,6 @@ def forward( >>> tokenizer.batch_decode(generate_ids, skip_special_tokens=True, clean_up_tokenization_spaces=False)[0] "Hey, are you conscious? Can you talk to me?\nI'm not conscious, but I can talk to you." ```""" - - output_hidden_states = ( - output_hidden_states if output_hidden_states is not None else self.config.output_hidden_states - ) # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) outputs: BaseModelOutputWithPast = self.model( @@ -1148,8 +1097,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, cache_position=cache_position, **kwargs, ) @@ -1171,8 +1118,6 @@ def forward( loss=loss, logits=logits, past_key_values=outputs.past_key_values, - hidden_states=outputs.hidden_states, - attentions=outputs.attentions, ) def prepare_inputs_for_generation( diff --git a/src/transformers/models/bart/modeling_bart.py b/src/transformers/models/bart/modeling_bart.py index a0a77668d6df..b3639c0238c0 100755 --- a/src/transformers/models/bart/modeling_bart.py +++ b/src/transformers/models/bart/modeling_bart.py @@ -193,7 +193,6 @@ def forward( past_key_value: Optional[Cache] = None, attention_mask: Optional[torch.Tensor] = None, layer_head_mask: Optional[torch.Tensor] = None, - output_attentions: bool = False, cache_position: Optional[torch.Tensor] = None, # TODO: we need a refactor so that the different attention modules can get their specific kwargs # ATM, we have mixed things encoder, decoder, and encoder-decoder attn @@ -259,7 +258,6 @@ def forward( attention_mask, dropout=0.0 if not self.training else self.dropout, scaling=self.scaling, - output_attentions=output_attentions, head_mask=layer_head_mask, **kwargs, ) @@ -295,7 +293,6 @@ def forward( hidden_states: torch.FloatTensor, attention_mask: torch.FloatTensor, layer_head_mask: torch.FloatTensor, - output_attentions: Optional[bool] = False, ) -> tuple[torch.FloatTensor, Optional[torch.FloatTensor]]: """ Args: @@ -333,12 +330,7 @@ def forward( clamp_value = torch.finfo(hidden_states.dtype).max - 1000 hidden_states = torch.clamp(hidden_states, min=-clamp_value, max=clamp_value) - outputs = (hidden_states,) - - if output_attentions: - outputs += (attn_weights,) - - return outputs + return hidden_states class BartDecoderLayer(nn.Module): @@ -449,15 +441,7 @@ def forward( hidden_states = residual + hidden_states hidden_states = self.final_layer_norm(hidden_states) - outputs = (hidden_states,) - - if output_attentions: - outputs += (self_attn_weights, cross_attn_weights) - - if use_cache: - outputs += (past_key_value,) - - return outputs + return hidden_states class BartClassificationHead(nn.Module): @@ -512,16 +496,6 @@ def _init_weights(self, module): module.weight.data.fill_(1.0) module.bias.data.zero_() - @property - def dummy_inputs(self): - pad_token = self.config.pad_token_id - input_ids = torch.tensor([[0, 6, 10, 4, 2], [0, 8, 12, 2, pad_token]], device=self.device) - dummy_inputs = { - "attention_mask": input_ids.ne(pad_token), - "input_ids": input_ids, - } - return dummy_inputs - def _update_full_mask( self, attention_mask: Union[torch.Tensor, None], @@ -779,8 +753,6 @@ def forward( attention_mask: Optional[torch.Tensor] = None, head_mask: Optional[torch.Tensor] = None, inputs_embeds: Optional[torch.FloatTensor] = None, - - return_dict: Optional[bool] = None, ) -> Union[tuple, BaseModelOutput]: r""" Args: @@ -818,21 +790,8 @@ def forward( return_dict (`bool`, *optional*): Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. """ - - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - - # retrieve input_ids and inputs_embeds - if input_ids is not None and inputs_embeds is not None: - raise ValueError("You cannot specify both input_ids and inputs_embeds at the same time") - elif input_ids is not None: - input = input_ids - input_ids = input_ids.view(-1, input_ids.shape[-1]) - elif inputs_embeds is not None: - input = inputs_embeds[:, :, -1] - else: - raise ValueError("You have to specify either input_ids or inputs_embeds") + # TODO just add the incompatibility between both input embeds and input_ids if inputs_embeds is None: inputs_embeds = self.embed_tokens(input_ids) @@ -848,21 +807,7 @@ def forward( inputs_embeds, ) - encoder_states = () if output_hidden_states else None - all_attentions = () if output_attentions else None - - # check if head_mask has a correct number of layers specified if desired - if head_mask is not None: - if head_mask.size()[0] != (len(self.layers)): - raise ValueError( - f"The head_mask should be specified for {len(self.layers)} layers, but it is for" - f" {head_mask.size()[0]}." - ) - for idx, encoder_layer in enumerate(self.layers): - if output_hidden_states: - encoder_states = encoder_states + (hidden_states,) - # add LayerDrop (see https://huggingface.co/papers/1909.11556 for description) to_drop = False if self.training: dropout_probability = torch.rand([]) @@ -872,35 +817,14 @@ def forward( if to_drop: layer_outputs = (None, None) else: - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - encoder_layer.__call__, - hidden_states, - attention_mask, - (head_mask[idx] if head_mask is not None else None), - output_attentions, - ) - else: - layer_outputs = encoder_layer( - hidden_states, - attention_mask, - layer_head_mask=(head_mask[idx] if head_mask is not None else None), - output_attentions=output_attentions, - ) + layer_outputs = encoder_layer( + hidden_states, + attention_mask, + layer_head_mask=(head_mask[idx] if head_mask is not None else None), + ) hidden_states = layer_outputs[0] - - if output_attentions: - all_attentions = all_attentions + (layer_outputs[1],) - - if output_hidden_states: - encoder_states = encoder_states + (hidden_states,) - - if not return_dict: - return tuple(v for v in [hidden_states, encoder_states, all_attentions] if v is not None) - return BaseModelOutput( - last_hidden_state=hidden_states, hidden_states=encoder_states, attentions=all_attentions - ) + return BaseModelOutput(last_hidden_state=hidden_states) class BartDecoder(BartPreTrainedModel): @@ -945,6 +869,7 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.embed_tokens = value + @check_model_inputs def forward( self, input_ids: Optional[torch.LongTensor] = None, @@ -956,105 +881,12 @@ def forward( past_key_values: Optional[list[torch.FloatTensor]] = None, inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, BaseModelOutputWithPastAndCrossAttentions]: - r""" - Args: - input_ids (`torch.LongTensor` of shape `(batch_size, sequence_length)`): - Indices of input sequence tokens in the vocabulary. Padding will be ignored by default should you - provide it. - - Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and - [`PreTrainedTokenizer.__call__`] for details. - - [What are input IDs?](../glossary#input-ids) - attention_mask (`torch.Tensor` of shape `(batch_size, sequence_length)`, *optional*): - Mask to avoid performing attention on padding token indices. Mask values selected in `[0, 1]`: - - - 1 for tokens that are **not masked**, - - 0 for tokens that are **masked**. - - [What are attention masks?](../glossary#attention-mask) - encoder_hidden_states (`torch.FloatTensor` of shape `(batch_size, encoder_sequence_length, hidden_size)`, *optional*): - Sequence of hidden-states at the output of the last layer of the encoder. Used in the cross-attention - of the decoder. - encoder_attention_mask (`torch.LongTensor` of shape `(batch_size, encoder_sequence_length)`, *optional*): - Mask to avoid performing cross-attention on padding tokens indices of encoder input_ids. Mask values - selected in `[0, 1]`: - - - 1 for tokens that are **not masked**, - - 0 for tokens that are **masked**. - - [What are attention masks?](../glossary#attention-mask) - head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*): - Mask to nullify selected heads of the attention modules. Mask values selected in `[0, 1]`: - - - 1 indicates the head is **not masked**, - - 0 indicates the head is **masked**. - - cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*): - Mask to nullify selected heads of the cross-attention modules in the decoder to avoid performing - cross-attention on hidden heads. Mask values selected in `[0, 1]`: - - - 1 indicates the head is **not masked**, - - 0 indicates the head is **masked**. - - past_key_values (`tuple(tuple(torch.FloatTensor))`, *optional*, returned when `use_cache=True` is passed or when `config.use_cache=True`): - Tuple of `tuple(torch.FloatTensor)` of length `config.n_layers`, with each tuple having 2 tensors of - shape `(batch_size, num_heads, sequence_length, embed_size_per_head)`) and 2 additional tensors of - shape `(batch_size, num_heads, encoder_sequence_length, embed_size_per_head)`. - - Contains pre-computed hidden-states (key and values in the self-attention blocks and in the - cross-attention blocks) that can be used (see `past_key_values` input) to speed up sequential decoding. - - If `past_key_values` are used, the user can optionally input only the last `decoder_input_ids` (those - that don't have their past key value states given to this model) of shape `(batch_size, 1)` instead of - all `decoder_input_ids` of shape `(batch_size, sequence_length)`. - inputs_embeds (`torch.FloatTensor` of shape `(batch_size, sequence_length, hidden_size)`, *optional*): - Optionally, instead of passing `input_ids` you can choose to directly pass an embedded representation. - This is useful if you want more control over how to convert `input_ids` indices into associated vectors - than the model's internal embedding lookup matrix. - output_attentions (`bool`, *optional*): - Whether or not to return the attentions tensors of all attention layers. See `attentions` under - returned tensors for more detail. - output_hidden_states (`bool`, *optional*): - Whether or not to return the hidden states of all layers. See `hidden_states` under returned tensors - for more detail. - return_dict (`bool`, *optional*): - Whether or not to return a [`~utils.ModelOutput`] instead of a plain tuple. - cache_position (`torch.LongTensor` of shape `(sequence_length)`, *optional*): - Indices depicting the position of the input sequence tokens in the sequence. It is used to update the - cache in the correct position and to infer the complete sequence length. - """ - - - use_cache = use_cache if use_cache is not None else self.config.use_cache - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - - if self.gradient_checkpointing and self.training: - if use_cache: - logger.warning_once( - "`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`..." - ) - use_cache = False - - # retrieve input_ids and inputs_embeds - if (input_ids is None) ^ (inputs_embeds is not None): - raise ValueError("You cannot specify both decoder_input_ids and decoder_inputs_embeds at the same time") - elif input_ids is not None: - input = input_ids - input_shape = input.shape - input_ids = input_ids.view(-1, input_shape[-1]) - elif inputs_embeds is not None: - input_shape = inputs_embeds.size()[:-1] - input = inputs_embeds[:, :, -1] - else: - raise ValueError("You have to specify either decoder_input_ids or decoder_inputs_embeds") - if inputs_embeds is None: inputs_embeds = self.embed_tokens(input) + input_shape = inputs_embeds.shape # initialize `past_key_values` return_legacy_cache = False @@ -1107,90 +939,27 @@ def forward( hidden_states = nn.functional.dropout(hidden_states, p=self.dropout, training=self.training) - # decoder layers - all_hidden_states = () if output_hidden_states else None - all_self_attns = () if output_attentions else None - all_cross_attentions = () if (output_attentions and encoder_hidden_states is not None) else None - next_decoder_cache = None - - # check if head_mask/cross_attn_head_mask has a correct number of layers specified if desired - for attn_mask, mask_name in zip([head_mask, cross_attn_head_mask], ["head_mask", "cross_attn_head_mask"]): - if attn_mask is not None: - if attn_mask.size()[0] != (len(self.layers)): - raise ValueError( - f"The `{mask_name}` should be specified for {len(self.layers)} layers, but it is for" - f" {head_mask.size()[0]}." - ) - for idx, decoder_layer in enumerate(self.layers): - # add LayerDrop (see https://huggingface.co/papers/1909.11556 for description) - if output_hidden_states: - all_hidden_states += (hidden_states,) if self.training: dropout_probability = torch.rand([]) if dropout_probability < self.layerdrop: continue - if self.gradient_checkpointing and self.training: - layer_outputs = self._gradient_checkpointing_func( - decoder_layer.__call__, - hidden_states, - attention_mask, - encoder_hidden_states, - encoder_attention_mask, - head_mask[idx] if head_mask is not None else None, - cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None, - None, - output_attentions, - use_cache, - cache_position, - ) - else: - layer_outputs = decoder_layer( - hidden_states, - attention_mask=attention_mask, - encoder_hidden_states=encoder_hidden_states, - encoder_attention_mask=encoder_attention_mask, - layer_head_mask=(head_mask[idx] if head_mask is not None else None), - cross_attn_layer_head_mask=( - cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None - ), - past_key_value=past_key_values, - output_attentions=output_attentions, - use_cache=use_cache, - cache_position=cache_position, - ) + layer_outputs = decoder_layer( + hidden_states, + attention_mask=attention_mask, + encoder_hidden_states=encoder_hidden_states, + encoder_attention_mask=encoder_attention_mask, + layer_head_mask=(head_mask[idx] if head_mask is not None else None), + cross_attn_layer_head_mask=(cross_attn_head_mask[idx] if cross_attn_head_mask is not None else None), + past_key_value=past_key_values, + use_cache=use_cache, + cache_position=cache_position, + ) hidden_states = layer_outputs[0] - if use_cache: - next_decoder_cache = layer_outputs[3 if output_attentions else 1] - - if output_attentions: - all_self_attns += (layer_outputs[1],) - - if encoder_hidden_states is not None: - all_cross_attentions += (layer_outputs[2],) - - # add hidden states from the last decoder layer - if output_hidden_states: - all_hidden_states += (hidden_states,) - - next_cache = next_decoder_cache if use_cache else None - if return_legacy_cache: - next_cache = past_key_values.to_legacy_cache() - - if not return_dict: - return tuple( - v - for v in [hidden_states, next_cache, all_hidden_states, all_self_attns, all_cross_attentions] - if v is not None - ) return BaseModelOutputWithPastAndCrossAttentions( last_hidden_state=hidden_states, - past_key_values=next_cache, - hidden_states=all_hidden_states, - attentions=all_self_attns, - cross_attentions=all_cross_attentions, ) @@ -1252,39 +1021,9 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqModelOutput]: - r""" - decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Indices of decoder input sequence tokens in the vocabulary. - - Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and - [`PreTrainedTokenizer.__call__`] for details. - - [What are decoder input IDs?](../glossary#decoder-input-ids) - - Bart uses the `eos_token_id` as the starting token for `decoder_input_ids` generation. If `past_key_values` - is used, optionally only the last `decoder_input_ids` have to be input (see `past_key_values`). - - For translation and summarization training, `decoder_input_ids` should be provided. If no - `decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to the right - for denoising pre-training following the paper. - decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also - be used by default. - - If you want to change padding behavior, you should read [`modeling_bart._prepare_decoder_attention_mask`] - and modify to your needs. See diagram 1 in [the paper](https://huggingface.co/papers/1910.13461) for more - information on the default strategy. - cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*): - Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in `[0, - 1]`: - - - 1 indicates the head is **not masked**, - - 0 indicates the head is **masked**. - """ # different to other models, Bart automatically creates decoder_input_ids from # input_ids if no decoder_input_ids are provided if decoder_input_ids is None and decoder_inputs_embeds is None: @@ -1299,29 +1038,13 @@ def forward( input_ids, self.config.pad_token_id, self.config.decoder_start_token_id ) - - - use_cache = use_cache if use_cache is not None else self.config.use_cache - return_dict = return_dict if return_dict is not None else self.config.use_return_dict - if encoder_outputs is None: encoder_outputs = self.encoder( input_ids=input_ids, attention_mask=attention_mask, head_mask=head_mask, inputs_embeds=inputs_embeds, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, ) - # If the user passed a tuple for encoder_outputs, we wrap it in a BaseModelOutput when return_dict=True - elif return_dict and not isinstance(encoder_outputs, BaseModelOutput): - encoder_outputs = BaseModelOutput( - last_hidden_state=encoder_outputs[0], - hidden_states=encoder_outputs[1] if len(encoder_outputs) > 1 else None, - attentions=encoder_outputs[2] if len(encoder_outputs) > 2 else None, - ) - # decoder outputs consists of (dec_features, past_key_value, dec_hidden, dec_attn) decoder_outputs = self.decoder( input_ids=decoder_input_ids, @@ -1333,24 +1056,15 @@ def forward( past_key_values=past_key_values, inputs_embeds=decoder_inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, ) - if not return_dict: - return decoder_outputs + encoder_outputs - return Seq2SeqModelOutput( last_hidden_state=decoder_outputs.last_hidden_state, past_key_values=decoder_outputs.past_key_values, decoder_hidden_states=decoder_outputs.hidden_states, decoder_attentions=decoder_outputs.attentions, cross_attentions=decoder_outputs.cross_attentions, - encoder_last_hidden_state=encoder_outputs.last_hidden_state, - encoder_hidden_states=encoder_outputs.hidden_states, - encoder_attentions=encoder_outputs.attentions, ) @@ -1422,43 +1136,9 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqLMOutput]: r""" - decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Indices of decoder input sequence tokens in the vocabulary. - - Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and - [`PreTrainedTokenizer.__call__`] for details. - - [What are decoder input IDs?](../glossary#decoder-input-ids) - - Bart uses the `eos_token_id` as the starting token for `decoder_input_ids` generation. If `past_key_values` - is used, optionally only the last `decoder_input_ids` have to be input (see `past_key_values`). - - For translation and summarization training, `decoder_input_ids` should be provided. If no - `decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to the right - for denoising pre-training following the paper. - decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also - be used by default. - - If you want to change padding behavior, you should read [`modeling_bart._prepare_decoder_attention_mask`] - and modify to your needs. See diagram 1 in [the paper](https://huggingface.co/papers/1910.13461) for more - information on the default strategy. - cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*): - Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in `[0, - 1]`: - - - 1 indicates the head is **not masked**, - - 0 indicates the head is **masked**. - labels (`torch.LongTensor` of shape `(batch_size, sequence_length)`, *optional*): - Labels for computing the masked language modeling loss. Indices should either be in `[0, ..., - config.vocab_size]` or -100 (see `input_ids` docstring). Tokens with indices set to `-100` are ignored - (masked), the loss is only computed for the tokens with labels in `[0, ..., config.vocab_size]`. - Example summarization: ```python @@ -1500,7 +1180,6 @@ def forward( ['not', 'good', 'healthy', 'great', 'very'] ``` """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None: if use_cache: @@ -1524,9 +1203,6 @@ def forward( inputs_embeds=inputs_embeds, decoder_inputs_embeds=decoder_inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, ) @@ -1538,11 +1214,6 @@ def forward( labels = labels.to(lm_logits.device) loss_fct = CrossEntropyLoss() masked_lm_loss = loss_fct(lm_logits.view(-1, self.config.vocab_size), labels.view(-1)) - - if not return_dict: - output = (lm_logits,) + outputs[1:] - return ((masked_lm_loss,) + output) if masked_lm_loss is not None else output - return Seq2SeqLMOutput( loss=masked_lm_loss, logits=lm_logits, @@ -1551,8 +1222,6 @@ def forward( decoder_attentions=outputs.decoder_attentions, cross_attentions=outputs.cross_attentions, encoder_last_hidden_state=outputs.encoder_last_hidden_state, - encoder_hidden_states=outputs.encoder_hidden_states, - encoder_attentions=outputs.encoder_attentions, ) def prepare_decoder_input_ids_from_labels(self, labels: torch.Tensor): @@ -1607,51 +1276,11 @@ def forward( decoder_inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqSequenceClassifierOutput]: - r""" - decoder_input_ids (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Indices of decoder input sequence tokens in the vocabulary. - - Indices can be obtained using [`AutoTokenizer`]. See [`PreTrainedTokenizer.encode`] and - [`PreTrainedTokenizer.__call__`] for details. - - [What are decoder input IDs?](../glossary#decoder-input-ids) - - Bart uses the `eos_token_id` as the starting token for `decoder_input_ids` generation. If `past_key_values` - is used, optionally only the last `decoder_input_ids` have to be input (see `past_key_values`). - - For translation and summarization training, `decoder_input_ids` should be provided. If no - `decoder_input_ids` is provided, the model will create this tensor by shifting the `input_ids` to the right - for denoising pre-training following the paper. - decoder_attention_mask (`torch.LongTensor` of shape `(batch_size, target_sequence_length)`, *optional*): - Default behavior: generate a tensor that ignores pad tokens in `decoder_input_ids`. Causal mask will also - be used by default. - - If you want to change padding behavior, you should read [`modeling_bart._prepare_decoder_attention_mask`] - and modify to your needs. See diagram 1 in [the paper](https://huggingface.co/papers/1910.13461) for more - information on the default strategy. - cross_attn_head_mask (`torch.Tensor` of shape `(decoder_layers, decoder_attention_heads)`, *optional*): - Mask to nullify selected heads of the cross-attention modules in the decoder. Mask values selected in `[0, - 1]`: - - - 1 indicates the head is **not masked**, - - 0 indicates the head is **masked**. - labels (`torch.LongTensor` of shape `(batch_size,)`, *optional*): - Labels for computing the sequence classification/regression loss. Indices should be in `[0, ..., - config.num_labels - 1]`. If `config.num_labels > 1` a classification loss is computed (Cross-Entropy). - """ - return_dict = return_dict if return_dict is not None else self.config.use_return_dict if labels is not None: use_cache = False - if input_ids is None and inputs_embeds is not None: - raise NotImplementedError( - f"Passing input embeddings is currently not supported for {self.__class__.__name__}" - ) - outputs = self.model( input_ids, attention_mask=attention_mask, @@ -1664,9 +1293,6 @@ def forward( inputs_embeds=inputs_embeds, decoder_inputs_embeds=decoder_inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, ) hidden_states = outputs[0] # last hidden state @@ -1682,41 +1308,14 @@ def forward( loss = None if labels is not None: - labels = labels.to(logits.device) - if self.config.problem_type is None: - if self.config.num_labels == 1: - self.config.problem_type = "regression" - elif self.config.num_labels > 1 and (labels.dtype == torch.long or labels.dtype == torch.int): - self.config.problem_type = "single_label_classification" - else: - self.config.problem_type = "multi_label_classification" - - if self.config.problem_type == "regression": - loss_fct = MSELoss() - if self.config.num_labels == 1: - loss = loss_fct(logits.squeeze(), labels.squeeze()) - else: - loss = loss_fct(logits, labels) - elif self.config.problem_type == "single_label_classification": - loss_fct = CrossEntropyLoss() - loss = loss_fct(logits.view(-1, self.config.num_labels), labels.view(-1)) - elif self.config.problem_type == "multi_label_classification": - loss_fct = BCEWithLogitsLoss() - loss = loss_fct(logits, labels) - if not return_dict: - output = (logits,) + outputs[1:] - return ((loss,) + output) if loss is not None else output + self.loss_func(input_ids, labels, **kwargs) return Seq2SeqSequenceClassifierOutput( loss=loss, logits=logits, past_key_values=outputs.past_key_values, decoder_hidden_states=outputs.decoder_hidden_states, - decoder_attentions=outputs.decoder_attentions, - cross_attentions=outputs.cross_attentions, encoder_last_hidden_state=outputs.encoder_last_hidden_state, - encoder_hidden_states=outputs.encoder_hidden_states, - encoder_attentions=outputs.encoder_attentions, ) @@ -1752,7 +1351,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, decoder_inputs_embeds: Optional[torch.FloatTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, Seq2SeqQuestionAnsweringModelOutput]: @@ -1801,9 +1399,6 @@ def forward( inputs_embeds=inputs_embeds, decoder_inputs_embeds=decoder_inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, - return_dict=return_dict, cache_position=cache_position, ) @@ -1846,9 +1441,7 @@ def forward( decoder_hidden_states=outputs.decoder_hidden_states, decoder_attentions=outputs.decoder_attentions, cross_attentions=outputs.cross_attentions, - encoder_last_hidden_state=outputs.encoder_last_hidden_state, encoder_hidden_states=outputs.encoder_hidden_states, - encoder_attentions=outputs.encoder_attentions, ) @@ -1917,7 +1510,6 @@ def forward( inputs_embeds: Optional[torch.FloatTensor] = None, labels: Optional[torch.LongTensor] = None, use_cache: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, ) -> Union[tuple, CausalLMOutputWithCrossAttentions]: @@ -1949,8 +1541,6 @@ def forward( True ```""" - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict # decoder outputs consists of (dec_features, layer_state, dec_hidden, dec_attn) @@ -1964,8 +1554,6 @@ def forward( past_key_values=past_key_values, inputs_embeds=inputs_embeds, use_cache=use_cache, - output_attentions=output_attentions, - output_hidden_states=output_hidden_states, return_dict=return_dict, cache_position=cache_position, ) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 7381b1e44928..9bd172542903 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -959,13 +959,9 @@ def check_model_inputs(func): @wraps(func) def wrapper(self, *args, **kwargs): - print(kwargs) - output_attentions = kwargs.get("output_attentions", self.config.output_attentions) - output_hidden_states = kwargs.get("output_hidden_states", self.config.output_hidden_states) use_cache = kwargs.get("use_cache", self.config.use_cache) return_dict = kwargs.pop("return_dict", self.config.use_return_dict) - # Safely extract common config-backed flags kwargs.setdefault("use_cache", use_cache) kwargs["return_dict"] = kwargs.pop("return_dict", return_dict) From b1351120d1a912e0c8f7241f8e3e9f06ebec10a5 Mon Sep 17 00:00:00 2001 From: Arthur Date: Mon, 23 Jun 2025 04:10:23 +0800 Subject: [PATCH 45/51] up --- src/transformers/models/aya_vision/modeling_aya_vision.py | 6 +----- 1 file changed, 1 insertion(+), 5 deletions(-) diff --git a/src/transformers/models/aya_vision/modeling_aya_vision.py b/src/transformers/models/aya_vision/modeling_aya_vision.py index 38b056303fb0..fb2403911c85 100644 --- a/src/transformers/models/aya_vision/modeling_aya_vision.py +++ b/src/transformers/models/aya_vision/modeling_aya_vision.py @@ -281,12 +281,9 @@ def forward( use_cache: Optional[bool] = None, output_attentions: Optional[bool] = None, output_hidden_states: Optional[bool] = None, - return_dict: Optional[bool] = None, cache_position: Optional[torch.LongTensor] = None, **kwargs: Unpack[FlashAttentionKwargs], ) -> Union[tuple, AyaVisionModelOutputWithPast]: - - return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer @@ -477,8 +474,7 @@ def forward( >>> gen_tokens = model.generate(**inputs, max_new_tokens=300, do_sample=True, temperature=0.3) >>> processor.tokenizer.decode(gen_tokens[0][inputs.input_ids.shape[1]:], skip_special_tokens=True) ```""" - - + return_dict = return_dict if return_dict is not None else self.config.use_return_dict vision_feature_layer = ( vision_feature_layer if vision_feature_layer is not None else self.config.vision_feature_layer From eb727c50c85c806e09b634417d6aa8bbc2761623 Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 11:48:09 +0200 Subject: [PATCH 46/51] push what it should look like for llama --- src/transformers/modeling_utils.py | 9 +++++---- src/transformers/models/llama/modeling_llama.py | 13 ++++++++----- 2 files changed, 13 insertions(+), 9 deletions(-) diff --git a/src/transformers/modeling_utils.py b/src/transformers/modeling_utils.py index bfd283d4eddf..ddbc6a339c7d 100644 --- a/src/transformers/modeling_utils.py +++ b/src/transformers/modeling_utils.py @@ -2046,10 +2046,11 @@ def __init__(self, config: PretrainedConfig, *inputs, **kwargs): self._keep_in_fp32_modules = copy.copy(self.__class__._keep_in_fp32_modules) self._no_split_modules = self._no_split_modules or [] - self._can_record_outputs: Dict[str, Tuple[nn.Module, int]] = { - "hidden_states": (GradientCheckpointingLayer, 0), - "attentions": (GradientCheckpointingLayer, 1), - } + _param_to_record = {} + for module in self.modules(): + if hasattr(module, "return_hooks"): + _param_to_record.update({module.return_hooks[0]: (module, module.return_hooks[1])}) + self._can_record_outputs: Dict[str, Tuple[nn.Module, int]] = _param_to_record def post_init(self): """ diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index ac6b4790f6d2..e9d2a0237a37 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -20,7 +20,6 @@ from typing import Callable, Optional, Union import torch -import torch.utils.checkpoint from torch import nn from ...activations import ACT2FN @@ -198,6 +197,8 @@ def eager_attention_forward( class LlamaAttention(nn.Module): """Multi-headed attention from 'Attention Is All You Need' paper""" + return_hooks = {"attention", 0} + def __init__(self, config: LlamaConfig, layer_idx: int): super().__init__() self.config = config @@ -266,6 +267,8 @@ def forward( class LlamaDecoderLayer(GradientCheckpointingLayer): + return_hooks = {"hidden_states", 0} + def __init__(self, config: LlamaConfig, layer_idx: int): super().__init__() self.hidden_size = config.hidden_size @@ -287,12 +290,12 @@ def forward( cache_position: Optional[torch.LongTensor] = None, position_embeddings: Optional[tuple[torch.Tensor, torch.Tensor]] = None, # necessary, but kept here for BC **kwargs: Unpack[FlashAttentionKwargs], - ) -> tuple[torch.FloatTensor, torch.Tensor]: + ) -> tuple[torch.Tensor]: residual = hidden_states hidden_states = self.input_layernorm(hidden_states) # Self Attention - hidden_states, self_attn_weights = self.self_attn( + hidden_states = self.self_attn( hidden_states=hidden_states, attention_mask=attention_mask, position_ids=position_ids, @@ -302,7 +305,7 @@ def forward( cache_position=cache_position, position_embeddings=position_embeddings, **kwargs, - ) + )[0] hidden_states = residual + hidden_states # Fully Connected @@ -310,7 +313,7 @@ def forward( hidden_states = self.post_attention_layernorm(hidden_states) hidden_states = self.mlp(hidden_states) hidden_states = residual + hidden_states - return hidden_states, self_attn_weights + return hidden_states @auto_docstring From 6c34e8a5433a80ae151a102298c62263f5174041 Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 11:52:20 +0200 Subject: [PATCH 47/51] small fix --- src/transformers/utils/generic.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 9bd172542903..9b30ddff2c1a 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -989,8 +989,8 @@ def capture_fn(module, input, output): return capture_fn capture_flags = self._can_record_outputs.keys() - recordable_keys = {k.replace("output_"): v for k, v in capture_flags.items()} - if any(capture_flags.values()): + recordable_keys = {f"output_{k}": kwargs.get(f"output_{k}", False) for k, v in capture_flags.items()} + if any(recordable_keys.values()): for _, layer in self.named_modules(): for key, (cls, idx) in self._can_record_outputs.items(): if capture_flags.get(key, False) and isinstance(layer, cls): From 81b2678251b01fd403c7b43b43f9cd2463a62665 Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 11:53:20 +0200 Subject: [PATCH 48/51] nits --- src/transformers/utils/generic.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 9b30ddff2c1a..93f398d0b141 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -989,7 +989,7 @@ def capture_fn(module, input, output): return capture_fn capture_flags = self._can_record_outputs.keys() - recordable_keys = {f"output_{k}": kwargs.get(f"output_{k}", False) for k, v in capture_flags.items()} + recordable_keys = {f"output_{k}": kwargs.get(f"output_{k}", False) for k in capture_flags} if any(recordable_keys.values()): for _, layer in self.named_modules(): for key, (cls, idx) in self._can_record_outputs.items(): From cfa8c407795e8457da4c57d81aa625cdf2a1fa24 Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 11:53:51 +0200 Subject: [PATCH 49/51] todo comment --- src/transformers/utils/generic.py | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index 93f398d0b141..c94549858dd1 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -991,7 +991,10 @@ def capture_fn(module, input, output): capture_flags = self._can_record_outputs.keys() recordable_keys = {f"output_{k}": kwargs.get(f"output_{k}", False) for k in capture_flags} if any(recordable_keys.values()): - for _, layer in self.named_modules(): + for ( + _, + layer, + ) in self.named_modules(): # pretty sure we gotta attache the hooks to an instance and not a class for key, (cls, idx) in self._can_record_outputs.items(): if capture_flags.get(key, False) and isinstance(layer, cls): hook_fn = make_capture_fn(key, idx) From e155ae851062ab1fd88422f81ab51dcab82c1adb Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 16:16:26 +0200 Subject: [PATCH 50/51] fix attentions --- src/transformers/models/llama/modeling_llama.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/src/transformers/models/llama/modeling_llama.py b/src/transformers/models/llama/modeling_llama.py index e9d2a0237a37..4a524c8ac682 100644 --- a/src/transformers/models/llama/modeling_llama.py +++ b/src/transformers/models/llama/modeling_llama.py @@ -197,7 +197,7 @@ def eager_attention_forward( class LlamaAttention(nn.Module): """Multi-headed attention from 'Attention Is All You Need' paper""" - return_hooks = {"attention", 0} + return_hooks = {"attentions", 1} def __init__(self, config: LlamaConfig, layer_idx: int): super().__init__() From a304370d2118592f0e7e526b09cbcfa07056a8d2 Mon Sep 17 00:00:00 2001 From: Arthur Date: Thu, 26 Jun 2025 16:19:27 +0200 Subject: [PATCH 51/51] cleanup --- src/transformers/utils/generic.py | 24 +++++------------------- 1 file changed, 5 insertions(+), 19 deletions(-) diff --git a/src/transformers/utils/generic.py b/src/transformers/utils/generic.py index c94549858dd1..8d84201a7de9 100644 --- a/src/transformers/utils/generic.py +++ b/src/transformers/utils/generic.py @@ -971,14 +971,9 @@ def wrapper(self, *args, **kwargs): bound.apply_defaults() all_args = bound.arguments - # Extract input_ids and inputs_embeds more robustly - input_ids = all_args.get("input_ids", None) - inputs_embeds = all_args.get("inputs_embeds", None) - # TODO check the inputs ids vs input embeds. - # probably a key string saying which input are not compatible with each other - - use_cache = kwargs["use_cache"] - past_key_values = kwargs.get("past_key_values", None) + # TODO @Lysandre add the head we have today about GC and training + # and all of the rest that is general transformers checking + hooks = [] collected_outputs = {} @@ -996,7 +991,7 @@ def capture_fn(module, input, output): layer, ) in self.named_modules(): # pretty sure we gotta attache the hooks to an instance and not a class for key, (cls, idx) in self._can_record_outputs.items(): - if capture_flags.get(key, False) and isinstance(layer, cls): + if capture_flags.get(key, getattr(self.config, key, False)) and isinstance(layer, cls): hook_fn = make_capture_fn(key, idx) hooks.append(register_hook_if_needed(layer, hook_fn)) @@ -1005,17 +1000,8 @@ def capture_fn(module, input, output): if h is not None: h.remove() - # Insert collected outputs for key in collected_outputs: - # Optional: append a final model-level output like `last_hidden_state` for hidden_states - # if needed. You can customize this as per key semantics. - outputs[key] = tuple(collected_outputs[key]) if collected_outputs[key] else None - - # Fill missing keys with None if not using return_dict - if return_dict is False: - for key in recordable_keys: - if key not in outputs: - outputs[key] = None + outputs[key] = collected_outputs[key] if return_dict is False: outputs = outputs.to_tuple()