Skip to content

GitLab

Unverified Commit db452760 authored by Xiaoyu Zhang's avatar Xiaoyu Zhang Committed by GitHub
Browse files

[ci] fix llama4 ci error (#5126)

parent 57f99608
Hide whitespace changes
Inline Side-by-side
Showing with 2 additions and 0 deletions
python/sglang/srt/layers/quantization/compressed_tensors/compressed_tensors_moe.py
View file @ db452760
...@@ -285,6 +285,7 @@ class CompressedTensorsW8A8Fp8MoEMethod(CompressedTensorsMoEMethod): ...@@ -285,6 +285,7 @@ class CompressedTensorsW8A8Fp8MoEMethod(CompressedTensorsMoEMethod):
activation: str = "silu", activation: str = "silu",
inplace: bool = True, inplace: bool = True,
no_combine: bool = False, no_combine: bool = False,
apply_router_weight_on_input: bool = False,
) -> torch.Tensor: ) -> torch.Tensor:
from sglang.srt.layers.moe.fused_moe_triton import fused_experts from sglang.srt.layers.moe.fused_moe_triton import fused_experts
from sglang.srt.layers.moe.topk import select_experts from sglang.srt.layers.moe.topk import select_experts
...@@ -314,6 +315,7 @@ class CompressedTensorsW8A8Fp8MoEMethod(CompressedTensorsMoEMethod): ...@@ -314,6 +315,7 @@ class CompressedTensorsW8A8Fp8MoEMethod(CompressedTensorsMoEMethod):
w2_scale=layer.w2_weight_scale, w2_scale=layer.w2_weight_scale,
a1_scale=layer.w13_input_scale, a1_scale=layer.w13_input_scale,
a2_scale=layer.w2_input_scale, a2_scale=layer.w2_input_scale,
apply_router_weight_on_input=apply_router_weight_on_input,
) )
......
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