Skip to content

update topk softmax #985

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
shihaobai merged 1 commit into from
Jul 24, 2025
Merged

update topk softmax #985

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
8 changes: 1 addition & 7 deletions lightllm/common/fused_moe/topk_select.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -41,19 +41,13 @@ def fused_topk(

topk_weights = torch.empty(M, topk, dtype=torch.float32, device=hidden_states.device)
topk_ids = torch.empty(M, topk, dtype=torch.int32, device=hidden_states.device)
token_expert_indicies = torch.empty(M, topk, dtype=torch.int32, device=hidden_states.device)

sgl_ops.topk_softmax(
topk_weights,
topk_ids,
token_expert_indicies,
gating_output.float(), # TODO(woosuk): Optimize this.
renormalize=renormalize,
)
Comment on lines 45 to 50
Copy link
Contributor

@gemini-code-assist gemini-code-assist bot Jul 24, 2025

Choose a reason for hiding this comment

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

high

The sgl_ops.topk_softmax function signature has changed, now including renormalize as an argument. The test file unit_tests/common/fused_moe/test_softmax_topk.py needs to be updated to reflect this change, otherwise the tests will fail with a TypeError.

Please update the unit tests in unit_tests/common/fused_moe/test_softmax_topk.py to use the new API.

del token_expert_indicies # Not used. Will be used in the future.

if renormalize:
topk_weights = topk_weights / topk_weights.sum(dim=-1, keepdim=True)

return topk_weights, topk_ids


Expand Down