Skip to content

strip padding from text latent cache for Ernie #1520

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
dxqb merged 1 commit into from
Jun 17, 2026
Merged

strip padding from text latent cache for Ernie #1520

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 15 additions & 1 deletion modules/dataLoader/ErnieBaseDataLoader.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -14,6 +14,8 @@
from mgds.pipelineModules.DecodeVAE import DecodeVAE
from mgds.pipelineModules.EncodeMistralText import EncodeMistralText
from mgds.pipelineModules.EncodeVAE import EncodeVAE
from mgds.pipelineModules.PadMaskedTokens import PadMaskedTokens
from mgds.pipelineModules.PruneMaskedTokens import PruneMaskedTokens
from mgds.pipelineModules.RescaleImageChannels import RescaleImageChannels
from mgds.pipelineModules.SampleVAEDistribution import SampleVAEDistribution
from mgds.pipelineModules.SaveImage import SaveImage
Expand All @@ -36,11 +38,16 @@ def _preparation_modules(self, config: TrainConfig, model: ErnieModel):
apply_thread_safe_forward(model.text_encoder) # workaround for transformers#42673, unclear if Mistral is affected
encode_prompt = EncodeMistralText(tokens_name='tokens', tokens_attention_mask_in_name='tokens_mask', hidden_state_out_name='text_encoder_hidden_state', tokens_attention_mask_out_name='tokens_mask',
text_encoder=model.text_encoder, hidden_state_output_index=HIDDEN_STATES_LAYER, autocast_contexts=[model.autocast_context], dtype=model.train_dtype.torch_dtype())
prune_masked_tokens = PruneMaskedTokens(tokens_name='tokens', tokens_mask_name='tokens_mask', hidden_state_name='text_encoder_hidden_state')

modules = [rescale_image, encode_image, image_sample]
if config.masked_training or config.model_type.has_mask_input():
modules.append(downscale_mask)
modules += [tokenize_prompt, encode_prompt]

if config.latent_caching:
modules.append(prune_masked_tokens)

return modules

def _cache_modules(self, config: TrainConfig, model: ErnieModel, model_setup: BaseErnieSetup):
Expand Down Expand Up @@ -71,6 +78,8 @@ def _cache_modules(self, config: TrainConfig, model: ErnieModel, model_setup: Ba
)

def _output_modules(self, config: TrainConfig, model: ErnieModel, model_setup: BaseErnieSetup):
pad_masked_tokens = PadMaskedTokens(tokens_name='tokens', tokens_mask_name='tokens_mask', hidden_state_name='text_encoder_hidden_state', max_length=PROMPT_MAX_LENGTH)

output_names = [
'image_path', 'latent_image',
'prompt',
Expand All @@ -84,7 +93,7 @@ def _output_modules(self, config: TrainConfig, model: ErnieModel, model_setup: B

output_names.append('text_encoder_hidden_state')

return self._output_modules_from_out_names(
output_module_list = self._output_modules_from_out_names(
model, model_setup,
output_names=output_names,
config=config,
Expand All @@ -94,6 +103,11 @@ def _output_modules(self, config: TrainConfig, model: ErnieModel, model_setup: B
train_dtype=model.train_dtype,
)

if config.latent_caching:
output_module_list = [pad_masked_tokens] + output_module_list

return output_module_list

def _debug_modules(self, config: TrainConfig, model: ErnieModel):
debug_dir = os.path.join(config.debug_dir, "dataloader")

Expand Down