Skip to content

[GLM4MoE] Set attention_softmax_in_fp32 and bf16 defaults in GLMMoEMo… #4314

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

Open
zhanghonggeng wants to merge 1 commit into
base: develop
Choose a base branch
from
Open

[GLM4MoE] Set attention_softmax_in_fp32 and bf16 defaults in GLMMoEMo… #4314

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
3 changes: 3 additions & 0 deletions paddleformers/transformers/glm4_moe/modeling.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -88,6 +88,9 @@ class GLMMoEModelProvider(GPTModelProvider):
router_aux_loss_coef: float = 0.001
moe_grouped_gemm: bool = False

attention_softmax_in_fp32: bool = True
Copy link
Copy Markdown
Collaborator

@From00 From00 Apr 20, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

这个是不是可以默认启用就行，不需要通过开关控制？

bf16: bool = True


def eager_attention_forward(
module: nn.Layer,
Expand Down
Loading