Skip to content

BugFix: qwen pp cannot send bool when using fuse loss #11156

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
swgu98 merged 7 commits into from
Nov 7, 2025
Merged

BugFix: qwen pp cannot send bool when using fuse loss #11156

Changes from all commits
Commits
Show all changes
7 commits
Select commit Hold shift + click to select a range
9fa53b6
feat(model): add support for Qwen model in RL PipelineParallel
Jason233333 Oct 10, 2025
190f0c2
feat(model): add support for QwenMoe model in RL PipelineParallel
Jason233333 Oct 20, 2025
7b74240
feat(model): add support for QwenMoe model in RL PipelineParallel
Jason233333 Oct 20, 2025
18b5c32
Merge branch 'develop' of github.com:Jason233333/PaddleNLP into develop
Jason233333 Oct 21, 2025
80e81c5
BugFix: qwen model sequence parallel can not get batch size
Jason233333 Oct 23, 2025
b4b93cd
Merge branch 'PaddlePaddle:develop' into develop
Jason233333 Oct 31, 2025
9f7b913
BugFix: qwen pp cannot send bool when using fuse loss
Jason233333 Oct 31, 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
3 changes: 2 additions & 1 deletion paddlenlp/transformers/qwen2/modeling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1486,7 +1486,8 @@ def __init__(self, config: Qwen2Config, embedding_weights=None, transpose_y=Fals
def forward(self, hidden_states, tensor_parallel_output=None, batch_size=None):
# add this for fused_head_and_loss_fn
if self.config.use_fused_head_and_loss_fn:
return hidden_states, self.weight, None, self.transpose_y
# return hidden_states, self.weight, None, self.transpose_y
return hidden_states, self.weight, None, None

if self.config.sequence_parallel:
hidden_states = GatherOp.apply(hidden_states)
Expand Down
Loading