minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[NPU] Determine the topk norm_type through scoring_func (#31107)

Browse Source 
Co-authored-by: iridiumine <42236072+iridiumine@users.noreply.github.com>
Co-authored-by: zhaozx-cn <59479021+zhaozx-cn@users.noreply.github.com>
Co-authored-by: sglang-npu-bot <sglangnpu@163.com>
This commit is contained in:
McZyWu
2026-07-15 10:26:44 +03:00
committed by GitHub
co-authored by iridiumine zhaozx-cn sglang-npu-bot
parent dec0836302
commit 241937af87
2 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files
python/sglang/srt/hardware_backend/npu/moe/topk.py
+1 -1
View File
@@ -97,7 +97,7 @@ def fused_topk_npu(
group_select_mode=(1 if use_grouped_topk else 0), group_select_mode=(1 if use_grouped_topk else 0),
renorm=0, renorm=0,
# 1 for sigmoid, 0 for softmax # 1 for sigmoid, 0 for softmax
norm_type=1, norm_type=(0 if topk_config.scoring_func == "softmax" else 1),
routed_scaling_factor=( routed_scaling_factor=(
topk_config.routed_scaling_factor topk_config.routed_scaling_factor
if topk_config.apply_routed_scaling_factor_on_output if topk_config.apply_routed_scaling_factor_on_output
python/sglang/srt/models/glm4_moe_lite.py
+1
View File
@@ -225,6 +225,7 @@ class Glm4MoeLiteSparseMoeBlock(nn.Module):
quant_config=quant_config, quant_config=quant_config,
routed_scaling_factor=self.routed_scaling_factor, routed_scaling_factor=self.routed_scaling_factor,
prefix=add_prefix("experts", prefix), prefix=add_prefix("experts", prefix),
**({"scoring_func": "sigmoid"}),
) )
self.topk = TopK( self.topk = TopK(