Skip to content

GitLab

Unverified Commit 92f5d0f0 authored by Kunshang Ji's avatar Kunshang Ji Committed by GitHub
Browse files

[XPU] fix mxfp4 activation type (#35691) 


Signed-off-by: default avatarKunshang Ji <kunshang.ji@intel.com>
parent a60985b0
Hide whitespace changes
Inline Side-by-side
Showing with 1 addition and 1 deletion
vllm/model_executor/layers/quantization/mxfp4.py
View file @ 92f5d0f0
...@@ -1258,7 +1258,7 @@ class XpuMxfp4MoEMethod(Mxfp4MoEMethod): ...@@ -1258,7 +1258,7 @@ class XpuMxfp4MoEMethod(Mxfp4MoEMethod):
topk_weights=routing_weights, topk_weights=routing_weights,
topk_ids=selected_experts, topk_ids=selected_experts,
n_experts_per_token=layer.top_k, n_experts_per_token=layer.top_k,
activation=layer.activation, activation=layer.activation.value,
num_experts=layer.local_num_experts, num_experts=layer.local_num_experts,
is_mxfp4=True, is_mxfp4=True,
) )
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