Skip to content

Update ppo_model_utils.py #10593

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 2 commits into from
May 19, 2025
Merged

Update ppo_model_utils.py #10593

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
54f70b1
Update ppo_model_utils.py
ZHUI May 14, 2025
f1a82de
Update pp_model_utils.py
ZHUI May 14, 2025
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
4 changes: 2 additions & 2 deletions paddlenlp/rl/models/pp_model_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,9 +38,9 @@
# Training patch
if self.training and self.is_pipeline_last_stage():
if getattr(self, "_step_losses", None):
self._step_losses.append(output.detach())
self._step_losses.append(output[0].detach())

Check warning on line 41 in paddlenlp/rl/models/pp_model_utils.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/rl/models/pp_model_utils.py#L41 

Added line #L41 was not covered by tests
else:
self._step_losses = [output.detach()]
self._step_losses = [output[0].detach()]

Check warning on line 43 in paddlenlp/rl/models/pp_model_utils.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/rl/models/pp_model_utils.py#L43 

Added line #L43 was not covered by tests


def make_wrapper(func, pre_patch=None, post_patch=None):
Expand Down
2 changes: 1 addition & 1 deletion paddlenlp/rl/models/ppo_model_utils.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -351,7 +351,7 @@


def entropy_from_logits(logits: paddle.Tensor, tensor_parallel_output=False):
return VocabParallelEntropy.apply(logits, tensor_parallel_output)
return VocabParallelEntropy.apply(logits.astype("float32"), tensor_parallel_output)

Check warning on line 354 in paddlenlp/rl/models/ppo_model_utils.py

View check run for this annotation

Codecov / codecov/patch

paddlenlp/rl/models/ppo_model_utils.py#L354 

Added line #L354 was not covered by tests


@merge_fwd_labels
Expand Down