Skip to content

Commit 645b04e

Browse files
committed
[Bugfix]: Fix missing SPLIT_K in GPTQ/AWQ MoE Triton config
Signed-off-by: kai.wang <[email protected]>
1 parent f72a817 commit 645b04e

File tree

1 file changed

+1
-0
lines changed

1 file changed

+1
-0
lines changed

vllm/model_executor/layers/fused_moe/fused_moe.py

Lines changed: 1 addition & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -614,6 +614,7 @@ def invoke_fused_moe_kernel(
614614
bit=4 if use_int4_w4a16 else 8,
615615
)
616616
config = config.copy()
617+
config["SPLIT_K"] = 1
617618
config.update(
618619
get_moe_wna16_block_config(
619620
config=config,

0 commit comments

Comments
 (0)