Skip to content

GitLab

"torchvision/vscode:/vscode.git/clone" did not exist on "fec39cd5d2e09bd5eda2b4b97b426708bd10b26d"
Unverified Commit 388c05d5 authored by yhyang201's avatar yhyang201 Committed by GitHub
Browse files

Fix bias handling in TritonMoeQuantInfo within quantization/mxfp4.py (#10579)

parent fc809665
Show whitespace changes
Inline Side-by-side
Showing with 2 additions and 2 deletions
python/sglang/srt/layers/quantization/mxfp4.py
View file @ 388c05d5
...@@ -731,8 +731,8 @@ class Mxfp4MoEMethod(FusedMoEMethodBase): ...@@ -731,8 +731,8 @@ class Mxfp4MoEMethod(FusedMoEMethodBase):
quant_info = TritonMoeQuantInfo( quant_info = TritonMoeQuantInfo(
w13_weight=layer.w13_weight, w13_weight=layer.w13_weight,
w2_weight=layer.w2_weight, w2_weight=layer.w2_weight,
w13_weight_bias=layer.w13_weight_bias, b13=getattr(layer, "w13_weight_bias", None),
w2_weight_bias=layer.w2_weight_bias, b2=getattr(layer, "w2_weight_bias", None),
) )
return self.runner.run(dispatch_output, quant_info) return self.runner.run(dispatch_output, quant_info)
......
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