Skip to content

[llm]fix pp no drop last #9439

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
merged 1 commit into from
Nov 19, 2024
Merged

[llm]fix pp no drop last #9439

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
9 changes: 9 additions & 0 deletions paddlenlp/trainer/trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3267,6 +3267,14 @@
else:
labels = None
inputs = inputs.pop("input_ids")
# consider no drop_last case
model_config_backup = model.micro_batch_size, model.accumulate_steps
if isinstance(inputs, tuple):
actual_batch_size = inputs[0].shape[0]
else:

Check warning on line 3274 in paddlenlp/trainer/trainer.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/trainer/trainer.py#L3272-L3274 

Added lines #L3272 - L3274 were not covered by tests
actual_batch_size = inputs.shape[0]
model.micro_batch_size = 1
model.accumulate_steps = actual_batch_size

Check warning on line 3277 in paddlenlp/trainer/trainer.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/trainer/trainer.py#L3276-L3277 

Added lines #L3276 - L3277 were not covered by tests

with paddle.no_grad():
if has_labels:
Expand All @@ -3276,6 +3284,7 @@
loss = loss.mean().detach()
else:
raise ValueError("pipeline mode eval need label!")
model.micro_batch_size, model.accumulate_steps = model_config_backup

Check warning on line 3287 in paddlenlp/trainer/trainer.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/trainer/trainer.py#L3287 

Added line #L3287 was not covered by tests

return (loss, None, labels)

Expand Down
Loading