Skip to content

[BugFix] Fix routed_scaling_factor_learnable not taking effect in cutlass backend apply_tp #7903

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
chang-wenbin merged 4 commits into from
May 25, 2026
+7 −6
Merged

[BugFix] Fix routed_scaling_factor_learnable not taking effect in cutlass backend apply_tp #7903

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
d3fec5d
fix_moe_learable-score111
chang-wenbin May 24, 2026
4273e4e
Merge branch 'develop' into fix_moe_learable-score111
chang-wenbin May 25, 2026
22baa59
Merge branch 'develop' into fix_moe_learable-score111
chang-wenbin May 25, 2026
8ab69cf
Merge branch 'develop' into fix_moe_learable-score111
plusNew001 May 25, 2026
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: 7 additions & 6 deletions fastdeploy/model_executor/layers/moe/fused_moe_cutlass_backend.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -446,7 +446,7 @@ def apply_tp(
gate_out = gate_out.cast("float32")
if fc1_latent_proj is not None:
x = fc1_latent_proj(x)
gate_out, topk_weights, topk_idx = get_moe_scores(
gate_out, _, __ = get_moe_scores(

This comment was marked as outdated.

Sign in to view

gate_out,
layer.n_group,
layer.topk_group,
Expand All @@ -458,11 +458,6 @@ def apply_tp(
use_fused_cast=use_fused,
)

if layer.routed_scaling_factor_learnable:
safe_topk_indices = paddle.clip(topk_idx, min=0)
gathered_scales = F.embedding(safe_topk_indices, layer.per_expert_scale.unsqueeze(1)).squeeze(-1)
topk_weights = topk_weights * gathered_scales

(
permute_input,
token_nums_per_expert,
Expand All @@ -484,6 +479,12 @@ def apply_tp(
self.moe_quant_type,
topk_only_mode=True,
)

if layer.routed_scaling_factor_learnable:
safe_topk_indices = paddle.clip(topk_idx, min=0)
gathered_scales = F.embedding(safe_topk_indices, layer.per_expert_scale.unsqueeze(1)).squeeze(-1)
topk_weights = topk_weights * gathered_scales
Copy link
Copy Markdown

@PaddlePaddle-bot PaddlePaddle-bot May 25, 2026

Choose a reason for hiding this comment

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

🟡 建议 MoE 多后端同步检查

当前修复只涉及 fused_moe_cutlass_backend.py，但 FastDeploy 存在多个 MoE backend 实现（fused_moe_deepgemm_backend.py、fused_moe_triton_backend.py、fused_moe_marlin_backend.py 等）。若这些 backend 中也存在类似的 routed_scaling_factor_learnable 逻辑，且同样放在 get_moe_scores 之后（而非 moe_expert_dispatch 之后），则可能存在相同 bug，建议排查并同步修复。


else:
gate_out = gate_out.cast("float32")
if fc1_latent_proj is not None:
Expand Down
Loading