Skip to content

GitLab

"...git@developer.sourcefind.cn:2222/OpenDAS/vllm_cscc.git" did not exist on "09e372e796a82da8bb24f1fcde60f804c023cf68"
Commit 66092265 authored by zhuwenwen's avatar zhuwenwen
Browse files

update N to N1

parent 531c6304
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/model_executor/layers/quantization/blockwise_int8.py
View file @ 66092265
...@@ -432,7 +432,7 @@ class BlockInt8MoEMethod: ...@@ -432,7 +432,7 @@ class BlockInt8MoEMethod:
E=layer.w13_weight.shape[0] E=layer.w13_weight.shape[0]
N1=layer.w13_weight.shape[1] N1=layer.w13_weight.shape[1]
N2=layer.w2_weight.shape[1] N2=layer.w2_weight.shape[1]
K=layer.w2_weight.shape[2] K=N1//2
if [E,N1,N2,K] not in self.tritonsingleton.moe_weight_shapes: if [E,N1,N2,K] not in self.tritonsingleton.moe_weight_shapes:
self.tritonsingleton.moe_weight_shapes.append([E,N1,N2,K]) self.tritonsingleton.moe_weight_shapes.append([E,N1,N2,K])
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment