Skip to content

feat: update grouped_topk to support softmax and sigmoid #3680

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
zhyncs merged 4 commits into from
Feb 21, 2025
Merged

feat: update grouped_topk to support softmax and sigmoid #3680

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
9ec7746
feat: update grouped_topk to support softmax and sigmoid
zixuanzhang226 Feb 20, 2025
885649c
Merge branch 'main' into feat/support-softmax-and-sigmoid-scoring-in-…
zixuanzhang226 Feb 20, 2025
4144cbc
Merge branch 'main' into feat/support-softmax-and-sigmoid-scoring-in-…
zixuanzhang226 Feb 20, 2025
daa0361
Merge branch 'main' into feat/support-softmax-and-sigmoid-scoring-in-…
zixuanzhang226 Feb 20, 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
13 changes: 10 additions & 3 deletions python/sglang/srt/layers/moe/topk.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,6 @@ def fused_topk(
return topk_weights, topk_ids


# This is used by the Deepseek V2/V3/R1 series models
@torch.compile(dynamic=True, backend=get_compiler_backend())
def grouped_topk(
hidden_states: torch.Tensor,
Expand All @@ -84,10 +83,17 @@ def grouped_topk(
renormalize: bool,
num_expert_group: int = 0,
topk_group: int = 0,
scoring_func: str = "softmax",
):
assert hidden_states.shape[0] == gating_output.shape[0], "Number of tokens mismatch"

scores = torch.softmax(gating_output, dim=-1)
if scoring_func == "softmax":
scores = torch.softmax(gating_output, dim=-1)
elif scoring_func == "sigmoid":
scores = gating_output.sigmoid()
else:
raise ValueError(f"Scoring function '{scoring_func}' is not supported.")

num_token = scores.shape[0]
group_scores = (
scores.view(num_token, num_expert_group, -1).max(dim=-1).values
Expand All @@ -111,6 +117,7 @@ def grouped_topk(
return topk_weights.to(torch.float32), topk_ids.to(torch.int32)


# DeepSeek V2/V3/R1 uses biased_grouped_top
@torch.compile(dynamic=True, backend=get_compiler_backend())
def biased_grouped_topk(
hidden_states: torch.Tensor,
Expand Down Expand Up @@ -165,7 +172,7 @@ def select_experts(
correction_bias: Optional[torch.Tensor] = None,
torch_native: bool = False,
):
# DeekSeekv2 uses grouped_top_k
# DeepSeek V2/V3/R1 uses biased_grouped_top
if use_grouped_topk:
assert topk_group is not None
assert num_expert_group is not None
Expand Down
Loading