From 9c908f05af7c346e6fc55cfd4d426904957ed3e0 Mon Sep 17 00:00:00 2001 From: Cyril Vallez Date: Tue, 27 Oct 2026 17:27:07 +0100 Subject: [PATCH 1/4] remove --- src/transformers/integrations/deepgemm.py | 3 -- .../models/hunyuan_vl/modeling_hunyuan_vl.py | 2 - .../models/idefics2/modeling_idefics2.py | 3 -- .../models/idefics3/modeling_idefics3.py | 3 -- .../modeling_llava_next_video.py | 3 -- .../modular_llava_next_video.py | 3 -- .../modeling_llava_onevision.py | 3 -- .../modular_llava_onevision.py | 2 - .../minimax_m3_vl/modeling_minimax_m3_vl.py | 2 - .../minimax_m3_vl/modular_minimax_m3_vl.py | 2 - .../models/mllama/modeling_mllama.py | 3 -- .../models/oneformer/modeling_oneformer.py | 2 - .../qwen2_5_omni/modeling_qwen2_5_omni.py | 9 ++--- .../qwen2_5_omni/modular_qwen2_5_omni.py | 8 +--- .../models/smolvlm/modeling_smolvlm.py | 3 -- .../models/smolvlm/modular_smolvlm.py | 2 - .../vibevoice_asr/modeling_vibevoice_asr.py | 37 +------------------ .../vibevoice_asr/modular_vibevoice_asr.py | 37 +------------------ 18 files changed, 9 insertions(+), 118 deletions(-) diff --git a/src/transformers/integrations/deepgemm.py b/src/transformers/integrations/deepgemm.py index 76810552213f..5a563873efd5 100644 --- a/src/transformers/integrations/deepgemm.py +++ b/src/transformers/integrations/deepgemm.py @@ -37,7 +37,6 @@ import torch from ..utils import logging -from ..utils.deprecation import deprecate_kwarg from ..utils.import_utils import ( is_kernels_available, maybe_import_error, @@ -577,14 +576,12 @@ def _combine_routed_output( # ── Public dispatches ────────────────────────────────────────────────────────── -@deprecate_kwarg("output_dtype", version="v5.16") def deepgemm_fp8_fp4_linear( input: torch.Tensor, weight: torch.Tensor, weight_scale_inv: torch.Tensor, bias: torch.Tensor | None = None, block_size: tuple[int, int] | None = None, - output_dtype: torch.dtype | None = None, activation_scale: torch.Tensor | None = None, ) -> torch.Tensor: """End-to-end DeepGEMM linear: per-token activation quant + FP8/FP4 matmul. diff --git a/src/transformers/models/hunyuan_vl/modeling_hunyuan_vl.py b/src/transformers/models/hunyuan_vl/modeling_hunyuan_vl.py index 02411963ff89..825b7a170168 100644 --- a/src/transformers/models/hunyuan_vl/modeling_hunyuan_vl.py +++ b/src/transformers/models/hunyuan_vl/modeling_hunyuan_vl.py @@ -37,7 +37,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import ( accepts_precomputed_kwargs, get_max_seqlen, @@ -352,7 +351,6 @@ def __init__(self, config: HunYuanVLVisionConfig): self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.embed_dim, bias=True) self.is_causal = False - @deprecate_kwarg("hidden_state", new_name="hidden_states", version="v5.20") def forward( self, hidden_states: torch.Tensor, diff --git a/src/transformers/models/idefics2/modeling_idefics2.py b/src/transformers/models/idefics2/modeling_idefics2.py index 22e56a1a2259..fd71afff7bb3 100644 --- a/src/transformers/models/idefics2/modeling_idefics2.py +++ b/src/transformers/models/idefics2/modeling_idefics2.py @@ -30,7 +30,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import merge_with_config_defaults from ...utils.output_capturing import capture_outputs from ..auto import AutoModel @@ -885,7 +884,6 @@ def get_image_features( image_batch_size would be 7 when num_images_per_sample=[1, 3, 1, 2] and max_num_images would be 3. """ ) - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, @@ -1018,7 +1016,6 @@ def get_image_features( @can_return_tuple @auto_docstring - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, diff --git a/src/transformers/models/idefics3/modeling_idefics3.py b/src/transformers/models/idefics3/modeling_idefics3.py index 34b6fdbe4df3..d3ee04c80878 100644 --- a/src/transformers/models/idefics3/modeling_idefics3.py +++ b/src/transformers/models/idefics3/modeling_idefics3.py @@ -29,7 +29,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import merge_with_config_defaults from ...utils.output_capturing import capture_outputs from ..auto import AutoModel @@ -627,7 +626,6 @@ def get_image_features( image_batch_size would be 7 when num_images_per_sample=[1, 3, 1, 2] and max_num_images would be 3. """ ) - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, @@ -760,7 +758,6 @@ def get_image_features( @can_return_tuple @auto_docstring - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, 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 fa4dace25561..e1c40799e1ad 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 @@ -35,7 +35,6 @@ from ...modeling_utils import PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ..auto import AutoModel from .configuration_llava_next_video import LlavaNextVideoConfig @@ -594,7 +593,6 @@ def forward( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring( custom_intro="Obtains video last hidden states from the vision tower and apply multimodal projection." ) @@ -821,7 +819,6 @@ def forward( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring def get_video_features( self, 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 9fa6129c5401..a6e876455ef9 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 @@ -26,7 +26,6 @@ from ...modeling_utils import slice_logits_to_keep from ...processing_utils import Unpack from ...utils import auto_docstring, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ..auto import AutoConfig from ..llava_next.modeling_llava_next import ( @@ -294,7 +293,6 @@ def get_image_features( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring( custom_intro="Obtains video last hidden states from the vision tower and apply multimodal projection." ) @@ -475,7 +473,6 @@ def forward( class LlavaNextVideoForConditionalGeneration(LlavaNextForConditionalGeneration): @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring def get_video_features( self, diff --git a/src/transformers/models/llava_onevision/modeling_llava_onevision.py b/src/transformers/models/llava_onevision/modeling_llava_onevision.py index c573b4fddd71..09c882805e34 100644 --- a/src/transformers/models/llava_onevision/modeling_llava_onevision.py +++ b/src/transformers/models/llava_onevision/modeling_llava_onevision.py @@ -35,7 +35,6 @@ from ...modeling_utils import PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ..auto import AutoModel from .configuration_llava_onevision import LlavaOnevisionConfig @@ -556,7 +555,6 @@ def forward( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring( custom_intro="Obtains video last hidden states from the vision tower, apply multimodal projection and pooling." ) @@ -785,7 +783,6 @@ def forward( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring def get_video_features( self, diff --git a/src/transformers/models/llava_onevision/modular_llava_onevision.py b/src/transformers/models/llava_onevision/modular_llava_onevision.py index 23f070934a8b..258a48bdf73c 100644 --- a/src/transformers/models/llava_onevision/modular_llava_onevision.py +++ b/src/transformers/models/llava_onevision/modular_llava_onevision.py @@ -37,7 +37,6 @@ from ...modeling_utils import slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TensorType, auto_docstring, logging -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ..llava_next.image_processing_llava_next import LlavaNextImageProcessor, LlavaNextImageProcessorKwargs from ..llava_next.image_processing_pil_llava_next import LlavaNextImageProcessorPil @@ -560,7 +559,6 @@ def get_image_features( @merge_with_config_defaults @can_return_tuple - @deprecate_kwarg("pixel_values", version="v5.22.0", new_name="pixel_values_videos") @auto_docstring( custom_intro="Obtains video last hidden states from the vision tower, apply multimodal projection and pooling." ) diff --git a/src/transformers/models/minimax_m3_vl/modeling_minimax_m3_vl.py b/src/transformers/models/minimax_m3_vl/modeling_minimax_m3_vl.py index 5ae40536bc96..d6cd17a8cb0d 100644 --- a/src/transformers/models/minimax_m3_vl/modeling_minimax_m3_vl.py +++ b/src/transformers/models/minimax_m3_vl/modeling_minimax_m3_vl.py @@ -42,7 +42,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import ModelOutput, TransformersKwargs, auto_docstring, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ...utils.import_utils import is_torchdynamo_compiling from ...utils.output_capturing import OutputRecorder, capture_outputs @@ -1174,7 +1173,6 @@ def __init__(self, config: MiniMaxM3VLVisionConfig): @merge_with_config_defaults @capture_outputs @auto_docstring - @deprecate_kwarg("image_grid_thw", new_name="grid_thw", version="5.16.0") def forward( self, pixel_values: torch.Tensor, grid_thw: torch.Tensor, **kwargs: Unpack[TransformersKwargs] ) -> BaseModelOutputWithPooling: diff --git a/src/transformers/models/minimax_m3_vl/modular_minimax_m3_vl.py b/src/transformers/models/minimax_m3_vl/modular_minimax_m3_vl.py index a5d826666ace..d75071683ca8 100644 --- a/src/transformers/models/minimax_m3_vl/modular_minimax_m3_vl.py +++ b/src/transformers/models/minimax_m3_vl/modular_minimax_m3_vl.py @@ -31,7 +31,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import can_return_tuple, merge_with_config_defaults from ...utils.import_utils import is_torchdynamo_compiling from ...utils.output_capturing import capture_outputs @@ -879,7 +878,6 @@ def __init__(self, config: MiniMaxM3VLVisionConfig): @merge_with_config_defaults @capture_outputs @auto_docstring - @deprecate_kwarg("image_grid_thw", new_name="grid_thw", version="5.16.0") def forward( self, pixel_values: torch.Tensor, grid_thw: torch.Tensor, **kwargs: Unpack[TransformersKwargs] ) -> BaseModelOutputWithPooling: diff --git a/src/transformers/models/mllama/modeling_mllama.py b/src/transformers/models/mllama/modeling_mllama.py index 07232d15ebbd..601c3f7922b6 100644 --- a/src/transformers/models/mllama/modeling_mllama.py +++ b/src/transformers/models/mllama/modeling_mllama.py @@ -35,7 +35,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import merge_with_config_defaults from ...utils.output_capturing import capture_outputs from .configuration_mllama import MllamaConfig, MllamaTextConfig, MllamaVisionConfig @@ -242,7 +241,6 @@ def __init__(self, config: MllamaVisionConfig): self.v_proj = nn.Linear(self.embed_dim, self.num_heads * self.head_dim, bias=False) self.o_proj = nn.Linear(self.num_heads * self.head_dim, self.embed_dim, bias=False) - @deprecate_kwarg("hidden_state", new_name="hidden_states", version="v5.20") def forward( self, hidden_states: torch.Tensor, @@ -300,7 +298,6 @@ def __init__(self, config: MllamaVisionConfig, is_gated: bool = False): self.gate_attn = nn.Parameter(torch.ones(1) * math.pi / 4) self.gate_ffn = nn.Parameter(torch.ones(1) * math.pi / 4) - @deprecate_kwarg("hidden_state", new_name="hidden_states", version="v5.20") def forward( self, hidden_states: torch.Tensor, diff --git a/src/transformers/models/oneformer/modeling_oneformer.py b/src/transformers/models/oneformer/modeling_oneformer.py index a88c258f2eac..7c928db66d4e 100644 --- a/src/transformers/models/oneformer/modeling_oneformer.py +++ b/src/transformers/models/oneformer/modeling_oneformer.py @@ -38,7 +38,6 @@ requires_backends, torch_compilable_check, ) -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import maybe_autocast from .configuration_oneformer import OneFormerConfig @@ -1152,7 +1151,6 @@ def __init__(self, config: OneFormerConfig): self.layers = nn.ModuleList([OneFormerPixelDecoderEncoderLayer(config) for _ in range(config.encoder_layers)]) @staticmethod - @deprecate_kwarg("spatial_shapes", version="5.22", new_name="spatial_shapes_list") def get_reference_points(spatial_shapes_list, valid_ratios, device): """ Get reference points for each feature map. Used in decoder. 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 03274a88fdd3..a89590cf66fe 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 @@ -55,7 +55,6 @@ logging, torch_compilable_check, ) -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import ( accepts_precomputed_kwargs, get_max_seqlen, @@ -1024,7 +1023,6 @@ def __init__(self, config: Qwen2_5OmniVisionEncoderConfig = None) -> None: self.attention_dropout = 0.0 self.is_causal = False - @deprecate_kwarg("rotary_pos_emb", version="v5.20", new_name="position_embeddings") def forward( self, hidden_states: torch.Tensor, @@ -1138,7 +1136,7 @@ def __init__(self, config: Qwen2_5OmniVisionEncoderConfig) -> None: self.attn = Qwen2_5OmniVisionAttention(config=config) self.mlp = Qwen2_5OmniMLP(config, bias=True) - @deprecate_kwarg("rotary_pos_emb", version="v5.20", new_name="position_embeddings") + @auto_docstring def forward( self, hidden_states: torch.Tensor, @@ -3820,7 +3818,6 @@ def from_pretrained( return model @torch.no_grad() - @deprecate_kwarg("return_audio", version="v5", new_name="generation_mode") # TODO: raushan, defaults should be saved in generation config def generate( self, @@ -3854,9 +3851,9 @@ def generate( - With a *thinker_*, *talker_*, *token2wav_* prefix, they will be input for the `generate` method of the thinker, talker and token2wav respectively. It has the priority over the keywords without a prefix. Returns: - When `return_audio=False`: + When `generation_mode="text"`: - **Text** (`torch.Tensor`): Generated text token sequence. - When `return_audio=True`: + Otherwise: - **Text** (`torch.Tensor`): Generated text token sequence. - **Audio waveform** (`torch.Tensor` or `list[torch.Tensor]`): A single `(num_samples,)` waveform when `batch_size == 1`, otherwise one waveform per batch sample, each trimmed to 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 9833e4fcfd4e..b82551fade49 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 @@ -47,7 +47,6 @@ logging, torch_compilable_check, ) -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import ( accepts_precomputed_kwargs, get_max_seqlen, @@ -1387,7 +1386,6 @@ def __init__(self, config: Qwen2_5OmniVisionEncoderConfig = None) -> None: self.attention_dropout = 0.0 self.is_causal = False - @deprecate_kwarg("rotary_pos_emb", version="v5.20", new_name="position_embeddings") def forward( self, hidden_states: torch.Tensor, @@ -1463,7 +1461,6 @@ def __init__(self, config: Qwen2_5OmniVisionEncoderConfig) -> None: super().__init__(config, config._attn_implementation) self.attn = Qwen2_5OmniVisionAttention(config=config) - @deprecate_kwarg("rotary_pos_emb", version="v5.20", new_name="position_embeddings") def forward( self, hidden_states: torch.Tensor, @@ -3663,7 +3660,6 @@ def from_pretrained( return model @torch.no_grad() - @deprecate_kwarg("return_audio", version="v5", new_name="generation_mode") # TODO: raushan, defaults should be saved in generation config def generate( self, @@ -3697,9 +3693,9 @@ def generate( - With a *thinker_*, *talker_*, *token2wav_* prefix, they will be input for the `generate` method of the thinker, talker and token2wav respectively. It has the priority over the keywords without a prefix. Returns: - When `return_audio=False`: + When `generation_mode="text"`: - **Text** (`torch.Tensor`): Generated text token sequence. - When `return_audio=True`: + Otherwise: - **Text** (`torch.Tensor`): Generated text token sequence. - **Audio waveform** (`torch.Tensor` or `list[torch.Tensor]`): A single `(num_samples,)` waveform when `batch_size == 1`, otherwise one waveform per batch sample, each trimmed to diff --git a/src/transformers/models/smolvlm/modeling_smolvlm.py b/src/transformers/models/smolvlm/modeling_smolvlm.py index 89fdd88d6e17..f91cfc541a26 100644 --- a/src/transformers/models/smolvlm/modeling_smolvlm.py +++ b/src/transformers/models/smolvlm/modeling_smolvlm.py @@ -36,7 +36,6 @@ from ...modeling_utils import ALL_ATTENTION_FUNCTIONS, PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import merge_with_config_defaults from ...utils.output_capturing import capture_outputs from ..auto import AutoModel @@ -530,7 +529,6 @@ def get_image_features( image_batch_size would be 7 when num_images_per_sample=[1, 3, 1, 2] and max_num_images would be 3. """ ) - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, @@ -704,7 +702,6 @@ def get_image_features( @can_return_tuple @auto_docstring - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, diff --git a/src/transformers/models/smolvlm/modular_smolvlm.py b/src/transformers/models/smolvlm/modular_smolvlm.py index 1dea95dbfa3f..a063beb04893 100644 --- a/src/transformers/models/smolvlm/modular_smolvlm.py +++ b/src/transformers/models/smolvlm/modular_smolvlm.py @@ -24,7 +24,6 @@ from ...modeling_outputs import BaseModelOutputWithPooling from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging, torch_compilable_check -from ...utils.deprecation import deprecate_kwarg from ...utils.generic import merge_with_config_defaults from ..idefics3.configuration_idefics3 import Idefics3Config, Idefics3VisionConfig from ..idefics3.image_processing_idefics3 import Idefics3ImageProcessor @@ -200,7 +199,6 @@ def get_image_features( image_batch_size would be 7 when num_images_per_sample=[1, 3, 1, 2] and max_num_images would be 3. """ ) - @deprecate_kwarg("image_hidden_states", version="v5.20", new_name="mm_encoder_outputs") def forward( self, input_ids: torch.LongTensor | None = None, diff --git a/src/transformers/models/vibevoice_asr/modeling_vibevoice_asr.py b/src/transformers/models/vibevoice_asr/modeling_vibevoice_asr.py index c4461e013caf..51055ada90d4 100644 --- a/src/transformers/models/vibevoice_asr/modeling_vibevoice_asr.py +++ b/src/transformers/models/vibevoice_asr/modeling_vibevoice_asr.py @@ -31,7 +31,6 @@ from ...modeling_utils import PreTrainedModel, slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, is_torchdynamo_compiling -from ...utils.deprecation import deprecate_kwarg from ..auto import AutoModel from .configuration_vibevoice_asr import VibeVoiceAsrConfig @@ -317,14 +316,12 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.language_model.set_input_embeddings(value) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring(custom_intro="Encode audio into embeddings that can be used by the language model.") def get_audio_features( self, input_values: torch.FloatTensor, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, **kwargs: Unpack[TransformersKwargs], ): r""" @@ -332,26 +329,12 @@ def get_audio_features( Input audio tensor sampled at 24kHz. padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks to process at once through the tokenizers. """ - if acoustic_tokenizer_chunk_size is None: - acoustic_tokenizer_chunk_size = self.config.acoustic_tokenizer_chunk_size - else: - if acoustic_tokenizer_chunk_size % self.config.acoustic_tokenizer_encoder_config.hop_length != 0: - acoustic_tokenizer_chunk_size = int( - (acoustic_tokenizer_chunk_size // self.config.acoustic_tokenizer_encoder_config.hop_length) - * self.config.acoustic_tokenizer_encoder_config.hop_length - ) - raise ValueError( - f"`acoustic_tokenizer_chunk_size` must be a multiple of hop length ({self.config.acoustic_tokenizer_encoder_config.hop_length}), {acoustic_tokenizer_chunk_size} is a valid option." - ) - with torch.no_grad(): acoustic_encoder_cache, semantic_encoder_cache = None, None acoustic_latents, semantic_latents = [], [] - for chunk in torch.split(input_values, acoustic_tokenizer_chunk_size, dim=-1): + for chunk in torch.split(input_values, self.config.acoustic_tokenizer_chunk_size, dim=-1): acoustic_encoder_output = self.acoustic_tokenizer_encoder( chunk, padding_cache=acoustic_encoder_cache, @@ -388,7 +371,6 @@ def get_audio_features( return BaseModelOutputWithPooling(last_hidden_state=acoustic_latents, pooler_output=combined_features) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring def forward( @@ -399,25 +381,18 @@ def forward( inputs_embeds: torch.FloatTensor | None = None, input_values: torch.FloatTensor | None = None, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, **kwargs: Unpack[TransformersKwargs], ) -> tuple | VibeVoiceAsrModelOutputWithPast: r""" padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks processed by the acoustic and semantic tokenizers. """ if inputs_embeds is None: inputs_embeds = self.get_input_embeddings()(input_ids) audio_embeds = None if input_values is not None and input_ids is not None: - audio_embeds = self.get_audio_features( - input_values=input_values, - padding_mask=padding_mask, - acoustic_tokenizer_chunk_size=acoustic_tokenizer_chunk_size, - ).pooler_output + audio_embeds = self.get_audio_features(input_values=input_values, padding_mask=padding_mask).pooler_output audio_token_mask = (input_ids == self.config.audio_token_id).unsqueeze(-1) inputs_embeds = inputs_embeds.masked_scatter( @@ -455,7 +430,6 @@ def __init__(self, config: VibeVoiceAsrConfig): def get_audio_features(self, *args, **kwargs): return self.model.get_audio_features(*args, **kwargs) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring def forward( @@ -466,7 +440,6 @@ def forward( inputs_embeds: torch.FloatTensor | None = None, input_values: torch.FloatTensor | None = None, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, labels: torch.LongTensor | None = None, logits_to_keep: int | torch.Tensor = 0, **kwargs: Unpack[TransformersKwargs], @@ -474,8 +447,6 @@ def forward( r""" padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks processed by the acoustic and semantic tokenizers. Example: @@ -493,7 +464,6 @@ def forward( inputs_embeds=inputs_embeds, input_values=input_values, padding_mask=padding_mask, - acoustic_tokenizer_chunk_size=acoustic_tokenizer_chunk_size, **kwargs, ) @@ -519,7 +489,6 @@ def forward( def prepare_inputs_for_generation(self, *args, is_first_iteration=False, **kwargs): input_values = kwargs.pop("input_values", None) padding_mask = kwargs.pop("padding_mask", None) - acoustic_tokenizer_chunk_size = kwargs.pop("acoustic_tokenizer_chunk_size", None) model_inputs = super().prepare_inputs_for_generation(*args, **kwargs) @@ -528,8 +497,6 @@ def prepare_inputs_for_generation(self, *args, is_first_iteration=False, **kwarg model_inputs["input_values"] = input_values if padding_mask is not None: model_inputs["padding_mask"] = padding_mask - if acoustic_tokenizer_chunk_size is not None: - model_inputs["acoustic_tokenizer_chunk_size"] = acoustic_tokenizer_chunk_size return model_inputs diff --git a/src/transformers/models/vibevoice_asr/modular_vibevoice_asr.py b/src/transformers/models/vibevoice_asr/modular_vibevoice_asr.py index 65257148b0a5..8e4ebd06a2f9 100644 --- a/src/transformers/models/vibevoice_asr/modular_vibevoice_asr.py +++ b/src/transformers/models/vibevoice_asr/modular_vibevoice_asr.py @@ -29,7 +29,6 @@ from ...modeling_utils import slice_logits_to_keep from ...processing_utils import Unpack from ...utils import TransformersKwargs, auto_docstring, can_return_tuple, logging -from ...utils.deprecation import deprecate_kwarg from ..auto import AutoConfig, AutoModel from ..qwen2.modeling_qwen2 import Qwen2RMSNorm from ..vibevoice_acoustic_tokenizer.modeling_vibevoice_acoustic_tokenizer import ( @@ -231,14 +230,12 @@ def get_input_embeddings(self): def set_input_embeddings(self, value): self.language_model.set_input_embeddings(value) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring(custom_intro="Encode audio into embeddings that can be used by the language model.") def get_audio_features( self, input_values: torch.FloatTensor, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, **kwargs: Unpack[TransformersKwargs], ): r""" @@ -246,26 +243,12 @@ def get_audio_features( Input audio tensor sampled at 24kHz. padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks to process at once through the tokenizers. """ - if acoustic_tokenizer_chunk_size is None: - acoustic_tokenizer_chunk_size = self.config.acoustic_tokenizer_chunk_size - else: - if acoustic_tokenizer_chunk_size % self.config.acoustic_tokenizer_encoder_config.hop_length != 0: - acoustic_tokenizer_chunk_size = int( - (acoustic_tokenizer_chunk_size // self.config.acoustic_tokenizer_encoder_config.hop_length) - * self.config.acoustic_tokenizer_encoder_config.hop_length - ) - raise ValueError( - f"`acoustic_tokenizer_chunk_size` must be a multiple of hop length ({self.config.acoustic_tokenizer_encoder_config.hop_length}), {acoustic_tokenizer_chunk_size} is a valid option." - ) - with torch.no_grad(): acoustic_encoder_cache, semantic_encoder_cache = None, None acoustic_latents, semantic_latents = [], [] - for chunk in torch.split(input_values, acoustic_tokenizer_chunk_size, dim=-1): + for chunk in torch.split(input_values, self.config.acoustic_tokenizer_chunk_size, dim=-1): acoustic_encoder_output = self.acoustic_tokenizer_encoder( chunk, padding_cache=acoustic_encoder_cache, @@ -302,7 +285,6 @@ def get_audio_features( return BaseModelOutputWithPooling(last_hidden_state=acoustic_latents, pooler_output=combined_features) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring def forward( @@ -313,25 +295,18 @@ def forward( inputs_embeds: torch.FloatTensor | None = None, input_values: torch.FloatTensor | None = None, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, **kwargs: Unpack[TransformersKwargs], ) -> tuple | VibeVoiceAsrModelOutputWithPast: r""" padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks processed by the acoustic and semantic tokenizers. """ if inputs_embeds is None: inputs_embeds = self.get_input_embeddings()(input_ids) audio_embeds = None if input_values is not None and input_ids is not None: - audio_embeds = self.get_audio_features( - input_values=input_values, - padding_mask=padding_mask, - acoustic_tokenizer_chunk_size=acoustic_tokenizer_chunk_size, - ).pooler_output + audio_embeds = self.get_audio_features(input_values=input_values, padding_mask=padding_mask).pooler_output audio_token_mask = (input_ids == self.config.audio_token_id).unsqueeze(-1) inputs_embeds = inputs_embeds.masked_scatter( @@ -369,7 +344,6 @@ def __init__(self, config: VibeVoiceAsrConfig): def get_audio_features(self, *args, **kwargs): return self.model.get_audio_features(*args, **kwargs) - @deprecate_kwarg("acoustic_tokenizer_chunk_size", version="v5.20") @can_return_tuple @auto_docstring def forward( @@ -380,7 +354,6 @@ def forward( inputs_embeds: torch.FloatTensor | None = None, input_values: torch.FloatTensor | None = None, padding_mask: torch.BoolTensor | None = None, - acoustic_tokenizer_chunk_size: int | None = None, labels: torch.LongTensor | None = None, logits_to_keep: int | torch.Tensor = 0, **kwargs: Unpack[TransformersKwargs], @@ -388,8 +361,6 @@ def forward( r""" padding_mask (`torch.BoolTensor` of shape `(batch_size, sequence_length)`, *optional*): Mask to avoid performing operations on padding feature indices. - acoustic_tokenizer_chunk_size (`int`, *optional*): - Size of audio chunks processed by the acoustic and semantic tokenizers. Example: @@ -407,7 +378,6 @@ def forward( inputs_embeds=inputs_embeds, input_values=input_values, padding_mask=padding_mask, - acoustic_tokenizer_chunk_size=acoustic_tokenizer_chunk_size, **kwargs, ) @@ -433,7 +403,6 @@ def forward( def prepare_inputs_for_generation(self, *args, is_first_iteration=False, **kwargs): input_values = kwargs.pop("input_values", None) padding_mask = kwargs.pop("padding_mask", None) - acoustic_tokenizer_chunk_size = kwargs.pop("acoustic_tokenizer_chunk_size", None) model_inputs = super().prepare_inputs_for_generation(*args, **kwargs) @@ -442,8 +411,6 @@ def prepare_inputs_for_generation(self, *args, is_first_iteration=False, **kwarg model_inputs["input_values"] = input_values if padding_mask is not None: model_inputs["padding_mask"] = padding_mask - if acoustic_tokenizer_chunk_size is not None: - model_inputs["acoustic_tokenizer_chunk_size"] = acoustic_tokenizer_chunk_size return model_inputs From dde25247942b9a5e28531d8b53d83f9a3546e2bf Mon Sep 17 00:00:00 2001 From: Cyril Vallez Date: Thu, 29 Oct 2026 16:15:08 +0100 Subject: [PATCH 2/4] fix --- tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) diff --git a/tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py b/tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py index 9dbf1438c2b9..fda39428476d 100644 --- a/tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py +++ b/tests/models/qwen2_5_omni/test_modeling_qwen2_5_omni.py @@ -685,7 +685,7 @@ def test_small_model_integration_test(self): inputs = inputs.to(torch_device) output = model.generate( - **inputs, thinker_temperature=0, thinker_do_sample=False, return_audio=False, thinker_max_new_tokens=20 + **inputs, thinker_temperature=0, thinker_do_sample=False, generation_mode="text", thinker_max_new_tokens=20 ) EXPECTED_DECODED_TEXT = Expectations({ @@ -710,7 +710,7 @@ def test_small_model_integration_test_batch(self): ).to(torch_device, dtype=torch.bfloat16) output = model.generate( - **inputs, thinker_temperature=0, thinker_do_sample=False, return_audio=False, thinker_max_new_tokens=20 + **inputs, thinker_temperature=0, thinker_do_sample=False, generation_mode="text", thinker_max_new_tokens=20 ) EXPECTED_DECODED_TEXTS = Expectations( @@ -758,7 +758,7 @@ def test_small_model_integration_test_multiturn(self): ).to(torch_device, dtype=torch.bfloat16) output = model.generate( - **inputs, thinker_temperature=0, thinker_do_sample=False, return_audio=False, thinker_max_new_tokens=20 + **inputs, thinker_temperature=0, thinker_do_sample=False, generation_mode="text", thinker_max_new_tokens=20 ) EXPECTED_DECODED_TEXT = "system\nYou are a helpful assistant.\nuser\nWhat's that sound and what kind of dog is this?\nassistant\nThe sound is glass shattering, and the dog appears to be a Labrador Retriever.\nuser\nHow about this one?\nassistant\nThe sound is a cough." @@ -1006,7 +1006,7 @@ def test_small_model_integration_test_batch_flashatt2(self): padding=True, ).to(torch_device) - output = model.generate(**inputs, thinker_temperature=0, thinker_do_sample=False, return_audio=False) + output = model.generate(**inputs, thinker_temperature=0, thinker_do_sample=False, generation_mode="text") EXPECTED_DECODED_TEXT = Expectations({ (None, None): "system\nYou are a helpful assistant.\nuser\nWhat's that sound and what kind of dog is this?\nassistant\nThe sound is glass shattering, and the dog appears to be a Labrador Retriever.", From 47fcb11719a4c665ca383f33350bd1ef54cb57e4 Mon Sep 17 00:00:00 2001 From: Cyril Vallez Date: Thu, 29 Oct 2026 16:27:23 +0100 Subject: [PATCH 3/4] remove the stuff that was deprecated but not removed...... --- .../models/mllama/modeling_mllama.py | 78 +++++++++---------- .../test_modeling_llava_onevision.py | 2 +- 2 files changed, 40 insertions(+), 40 deletions(-) diff --git a/src/transformers/models/mllama/modeling_mllama.py b/src/transformers/models/mllama/modeling_mllama.py index 601c3f7922b6..26e32ffad93e 100644 --- a/src/transformers/models/mllama/modeling_mllama.py +++ b/src/transformers/models/mllama/modeling_mllama.py @@ -122,15 +122,15 @@ def __init__(self, config: MllamaVisionConfig, is_gated: bool = True): if is_gated: self.gate = nn.Parameter(torch.zeros(1)) - def forward(self, hidden_state: torch.Tensor, aspect_ratio_ids: torch.Tensor) -> torch.Tensor: + def forward(self, hidden_states: torch.Tensor, aspect_ratio_ids: torch.Tensor) -> torch.Tensor: embeddings = self.embedding(aspect_ratio_ids) embeddings = embeddings.reshape(-1, self.max_num_tiles, 1, self.hidden_size) if self.is_gated: embeddings = embeddings * self.gate.tanh() - hidden_state = hidden_state + embeddings - return hidden_state + hidden_states = hidden_states + embeddings + return hidden_states class MllamaPrecomputedPositionEmbedding(nn.Module): @@ -153,21 +153,21 @@ def __init__(self, config: MllamaVisionConfig): self.max_aspect_ratio_id + 1, self.max_num_tiles * self.num_patches * self.hidden_size ) - def forward(self, hidden_state: torch.Tensor, aspect_ratio_ids: torch.Tensor) -> torch.Tensor: + def forward(self, hidden_states: torch.Tensor, aspect_ratio_ids: torch.Tensor) -> torch.Tensor: # position embeddings gated_position_embedding = (1 - self.gate.tanh()) * self.embedding - hidden_state = hidden_state + gated_position_embedding.view(1, 1, self.num_patches, self.hidden_size) + hidden_states = hidden_states + gated_position_embedding.view(1, 1, self.num_patches, self.hidden_size) # precomputed tile position embeddings tile_position_embedding = self.tile_embedding(aspect_ratio_ids) - batch_size = hidden_state.shape[0] + batch_size = hidden_states.shape[0] tile_position_embedding = tile_position_embedding.reshape( batch_size, self.max_num_tiles, self.num_patches, self.hidden_size ) gated_tile_position_embedding = self.gate.tanh() * tile_position_embedding - hidden_state = hidden_state + gated_tile_position_embedding + hidden_states = hidden_states + gated_tile_position_embedding - return hidden_state + return hidden_states # Copied from transformers.models.clip.modeling_clip.CLIPMLP with CLIP->MllamaVision @@ -361,7 +361,7 @@ def forward( encoder_states = () for encoder_layer in self.layers: hidden_states = encoder_layer( - hidden_state=hidden_states, + hidden_states=hidden_states, attention_mask=attention_mask, ) encoder_states = encoder_states + (hidden_states,) @@ -856,11 +856,11 @@ def get_input_embeddings(self): """ return self.patch_embedding - def apply_class_embedding(self, hidden_state: torch.Tensor) -> torch.Tensor: - batch_size, _, hidden_size = hidden_state.shape + def apply_class_embedding(self, hidden_states: torch.Tensor) -> torch.Tensor: + batch_size, _, hidden_size = hidden_states.shape class_embedding = self.class_embedding.expand(batch_size, 1, hidden_size) - hidden_state = torch.cat([class_embedding, hidden_state], dim=1) - return hidden_state + hidden_states = torch.cat([class_embedding, hidden_states], dim=1) + return hidden_states @merge_with_config_defaults @capture_outputs @@ -919,30 +919,30 @@ def forward( target_dtype = self.patch_embedding.weight.dtype target_device = self.patch_embedding.weight.device patch_embeds = self.patch_embedding(pixel_values.to(target_device, target_dtype)) - hidden_state = patch_embeds.flatten(2).transpose(1, 2) + hidden_states = patch_embeds.flatten(2).transpose(1, 2) # Tile embeddings - _, num_patches, dim = hidden_state.shape - hidden_state = hidden_state.reshape(batch_size * num_concurrent_media, num_tiles, -1, dim) - hidden_state = self.pre_tile_positional_embedding(hidden_state, aspect_ratio_ids) + _, num_patches, dim = hidden_states.shape + hidden_states = hidden_states.reshape(batch_size * num_concurrent_media, num_tiles, -1, dim) + hidden_states = self.pre_tile_positional_embedding(hidden_states, aspect_ratio_ids) # Add cls token - hidden_state = hidden_state.reshape(batch_size * num_concurrent_media * num_tiles, num_patches, dim) - hidden_state = self.apply_class_embedding(hidden_state) + hidden_states = hidden_states.reshape(batch_size * num_concurrent_media * num_tiles, num_patches, dim) + hidden_states = self.apply_class_embedding(hidden_states) num_patches += 1 # Position embeddings - hidden_state = hidden_state.reshape(batch_size * num_concurrent_media, num_tiles, num_patches, dim) - hidden_state = self.gated_positional_embedding(hidden_state, aspect_ratio_ids) + hidden_states = hidden_states.reshape(batch_size * num_concurrent_media, num_tiles, num_patches, dim) + hidden_states = self.gated_positional_embedding(hidden_states, aspect_ratio_ids) - hidden_state = self.layernorm_pre(hidden_state) + hidden_states = self.layernorm_pre(hidden_states) # Compute the number of tokens to pad - num_padding_patches = (8 - (hidden_state.shape[-2] % 8)) % 8 + num_padding_patches = (8 - (hidden_states.shape[-2] % 8)) % 8 # Compute padding tuple for pad function padding = (0, 0, 0, num_padding_patches) # (pad_left, pad_right, pad_left for dim -2, pad_right for dim -2) # Pad the tensor - hidden_state = F.pad(hidden_state, padding, mode="constant", value=0) + hidden_states = F.pad(hidden_states, padding, mode="constant", value=0) slice_index = -num_padding_patches if num_padding_patches > 0 else None # Prepare attention mask @@ -950,40 +950,40 @@ def forward( attention_mask = _prepare_aspect_ratio_attention_mask( aspect_ratio_mask=attention_mask, num_patches=self.num_patches, - target_length=hidden_state.shape[2], + target_length=hidden_states.shape[2], dtype=self.dtype, ) # Apply encoder - hidden_state = hidden_state.view(batch_size * num_concurrent_media, -1, dim) + hidden_states = hidden_states.view(batch_size * num_concurrent_media, -1, dim) output = self.transformer( - hidden_state, + hidden_states, attention_mask=attention_mask, ) - hidden_state = output.last_hidden_state + hidden_states = output.last_hidden_state - hidden_state = self.layernorm_post(hidden_state) + hidden_states = self.layernorm_post(hidden_states) # Apply global encoder - hidden_state = hidden_state.reshape( + hidden_states = hidden_states.reshape( batch_size * num_concurrent_media, num_tiles, num_patches + num_padding_patches, dim ) - hidden_state = self.post_tile_positional_embedding(hidden_state, aspect_ratio_ids) - hidden_state = hidden_state.reshape( + hidden_states = self.post_tile_positional_embedding(hidden_states, aspect_ratio_ids) + hidden_states = hidden_states.reshape( batch_size * num_concurrent_media, num_tiles * (num_patches + num_padding_patches), dim ) global_output = self.global_transformer( - hidden_state, + hidden_states, attention_mask=attention_mask, ) - hidden_state = global_output.last_hidden_state + hidden_states = global_output.last_hidden_state # Remove padding form hidden state - hidden_state = hidden_state.reshape( + hidden_states = hidden_states.reshape( batch_size * num_concurrent_media, num_tiles, num_patches + num_padding_patches, dim ) - hidden_state = hidden_state[:, :, :slice_index] - hidden_state = hidden_state.reshape(batch_size, num_concurrent_media, num_tiles, num_patches, dim) + hidden_states = hidden_states[:, :, :slice_index] + hidden_states = hidden_states.reshape(batch_size, num_concurrent_media, num_tiles, num_patches, dim) # Collect intermediate layer outputs from encoder output all_intermediate_hidden_states = [output.hidden_states[i] for i in self.intermediate_layers_indices] @@ -999,9 +999,9 @@ def forward( ) # Concatenate final hidden state and intermediate hidden states - hidden_state = torch.cat([hidden_state, intermediate_hidden_states], dim=-1) + hidden_states = torch.cat([hidden_states, intermediate_hidden_states], dim=-1) - return BaseModelOutput(last_hidden_state=hidden_state) + return BaseModelOutput(last_hidden_state=hidden_states) @auto_docstring( diff --git a/tests/models/llava_onevision/test_modeling_llava_onevision.py b/tests/models/llava_onevision/test_modeling_llava_onevision.py index 3594da48e5b8..dfaa78ed81d6 100644 --- a/tests/models/llava_onevision/test_modeling_llava_onevision.py +++ b/tests/models/llava_onevision/test_modeling_llava_onevision.py @@ -308,7 +308,7 @@ def _video_features_prepare_config_and_inputs(self): ] ) config = self.model_tester.get_config() - inputs_dict = {"pixel_values": pixel_values_videos} + inputs_dict = {"pixel_values_videos": pixel_values_videos} return config, inputs_dict From 0ec6b7c140113adbf91241f545309473262dbb82 Mon Sep 17 00:00:00 2001 From: Cyril Vallez Date: Thu, 29 Oct 2026 16:45:23 +0100 Subject: [PATCH 4/4] fix more --- .../test_modeling_llava_next_video.py | 11 ----------- 1 file changed, 11 deletions(-) diff --git a/tests/models/llava_next_video/test_modeling_llava_next_video.py b/tests/models/llava_next_video/test_modeling_llava_next_video.py index 97e844c816a5..8bc7775e885c 100644 --- a/tests/models/llava_next_video/test_modeling_llava_next_video.py +++ b/tests/models/llava_next_video/test_modeling_llava_next_video.py @@ -328,17 +328,6 @@ def test_flash_attn_2_fp32_ln(self): def test_flash_attention_2_padding_matches_padding_free_with_position_ids(self): pass - def _video_features_prepare_config_and_inputs(self): - """ - Helper method to extract only image-related inputs from the full set of inputs, for testing `get_image_features`. - - Despite using `pixel_values_videos` in forward, LlavaNextVideo's `get_video_features` method - instead uses `pixel_values` as input, so we need to override the inputs accordingly. - """ - config, inputs_dict = self.model_tester.prepare_config_and_inputs_for_common() - inputs_dict = {"pixel_values": inputs_dict["pixel_values_videos"]} - return config, inputs_dict - @require_torch class LlavaNextVideoForConditionalGenerationIntegrationTest(unittest.TestCase):