apps/FastDeploy
1
0
Fork 0
mirror of https://github.com/PaddlePaddle/FastDeploy.git synced 2025-10-05 08:37:06 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

fix glm all_reduce tp group (#4188)

Browse Source
This commit is contained in:
chen
2025-09-22 10:57:13 +08:00
committed by GitHub
parent bba279cf38
commit 7bdc6f41e5
1 changed files with 1 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
fastdeploy/model_executor/models/glm4_moe.py
View File

@@ -160,7 +160,7 @@ class Glm4Moe(nn.Layer):
out = out + shared_experts_out out = out + shared_experts_out
# We do to TP all reduce after the sum of experts. # We do to TP all reduce after the sum of experts.
if self.tensor_parallel_size > 1: if self.tensor_parallel_size > 1:
tensor_model_parallel_all_reduce(out) tensor_model_parallel_all_reduce(out, self.tp_group)
return out return out