EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[NVIDIA]Fix local_num_experts for EP (#8779)

Browse Source
This commit is contained in:
Shu Wang
2025-08-05 00:01:14 -05:00
committed by GitHub
parent 1ea94d3b92
commit b01eeb80f8
2 changed files with 4 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
python/sglang/srt/layers/moe/fused_moe_triton/layer.py
View File

@@ -200,7 +200,8 @@ class FusedMoE(torch.nn.Module):
self.quant_config = quant_config
self.quant_method.create_weights(
layer=self,
num_experts=self.num_local_experts,
num_experts=self.num_experts,
num_local_experts=self.num_local_experts,
hidden_size=hidden_size,
# FIXME: figure out which intermediate_size to use
intermediate_size=self.intermediate_size_per_partition,

3
python/sglang/srt/layers/quantization/modelopt_quant.py
View File

@@ -752,6 +752,7 @@ class ModelOptNvFp4FusedMoEMethod(FusedMoEMethodBase):
self,
layer: torch.nn.Module,
num_experts: int,
num_local_experts: int,
hidden_size: int,
intermediate_size_per_partition: int,
params_dtype: torch.dtype,
@@ -765,7 +766,7 @@ class ModelOptNvFp4FusedMoEMethod(FusedMoEMethodBase):
# TODO(ch-wan): check if this is needed
layer.num_experts = num_experts
layer.num_local_experts = num_experts
layer.num_local_experts = num_local_experts
layer.intermediate_size_per_partition = intermediate_size_per_partition
layer.params_dtype = params_dtype
layer.quant_config = self.quant_config