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

[Fix] fix type issue of env flag value MODELOPT_MAX_TOKENS_PER_EXPERT (#11709)

Browse Source 
Signed-off-by: zejunchen-zejun <zejun.chen@amd.com>
This commit is contained in:
zejunchen-zejun
2025-10-30 00:44:05 +08:00
committed by GitHub
parent 52694b60da
commit 8a6838212a
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
sgl-kernel/python/sgl_kernel/gemm.py
View File

@@ -465,7 +465,7 @@ def scaled_fp4_experts_quant(
# larger models.
import os
MAX_TOKENS_PER_EXPERT = os.environ.get("MODELOPT_MAX_TOKENS_PER_EXPERT", 65536)
MAX_TOKENS_PER_EXPERT = int(os.environ.get("MODELOPT_MAX_TOKENS_PER_EXPERT", 65536))
assert m_numtopk <= MAX_TOKENS_PER_EXPERT * topk, (
f"m_numtopk must be less than MAX_TOKENS_PER_EXPERT("
f"{MAX_TOKENS_PER_EXPERT})"