biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Feat] Refactor for parallel_config in FusedMoEModularKernel (#30282)

Browse Source 
Signed-off-by: yewentao256 <zhyanwentao@126.com>
Signed-off-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <robshaw@redhat.com>
Co-authored-by: Robert Shaw <114415538+robertgshaw2-redhat@users.noreply.github.com>
This commit is contained in:
Wentao Ye
2025-12-14 23:21:36 -05:00
committed by GitHub
parent b337647aa0
commit 3778673ea8
8 changed files with 32 additions and 27 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm/model_executor/layers/quantization/utils/flashinfer_utils.py
View File

@@ -247,11 +247,6 @@ def flashinfer_cutlass_moe_fp8(
assert quant_config is not None
# Construct modular kernel with block-scale support when requested.
parallel_config = getattr(
getattr(layer, "vllm_config", None),
"parallel_config",
None,
)
fused_experts = mk.FusedMoEModularKernel(
build_flashinfer_fp8_cutlass_moe_prepare_finalize(
moe=moe, use_deepseek_fp8_block_scale=use_deepseek_fp8_block_scale
@@ -262,7 +257,7 @@ def flashinfer_cutlass_moe_fp8(
out_dtype=hidden_states.dtype,
use_deepseek_fp8_block_scale=use_deepseek_fp8_block_scale,
),
parallel_config=parallel_config,
moe_parallel_config=layer.moe_parallel_config,
)
return fused_experts(