[MoE] Move GPT OSS Triton kernel experts into fused_moe/experts/ (#39007)

Signed-off-by: Robert Shaw <robertgshaw2@gmail.com>
Signed-off-by: Jackmin801 <ongjackm@gmail.com>
Co-authored-by: Robert Shaw <robertgshaw2@gmail.com>
This commit is contained in:
Jackmin801
2026-04-14 19:27:39 +00:00
committed by GitHub
co-authored by Robert Shaw
parent ecf5ff7ce3
commit 1a9353bb02
8 changed files with 16 additions and 12 deletions
@@ -101,7 +101,7 @@ def backend_to_kernel_cls(
return [FlashInferExperts]
elif backend == Mxfp4MoeBackend.TRITON:
from vllm.model_executor.layers.fused_moe.gpt_oss_triton_kernels_moe import (
from vllm.model_executor.layers.fused_moe.experts.gpt_oss_triton_kernels_moe import ( # noqa: E501
OAITritonExperts,
OAITritonMxfp4ExpertsMonolithic,
)
@@ -110,7 +110,7 @@ def backend_to_kernel_cls(
return [OAITritonMxfp4ExpertsMonolithic, OAITritonExperts]
elif backend == Mxfp4MoeBackend.TRITON_UNFUSED:
from vllm.model_executor.layers.fused_moe.gpt_oss_triton_kernels_moe import (
from vllm.model_executor.layers.fused_moe.experts.gpt_oss_triton_kernels_moe import ( # noqa: E501
UnfusedOAITritonExperts,
)
@@ -1591,7 +1591,7 @@ class QuarkOCP_MX_MoEMethod_OSS(QuarkOCP_MX_MoEMethod):
"EPLB not supported for `QuarkW4MXFp4MoEMethod_OSS` yet."
)
from vllm.model_executor.layers.fused_moe.gpt_oss_triton_kernels_moe import ( # noqa: E501
from vllm.model_executor.layers.fused_moe.experts.gpt_oss_triton_kernels_moe import ( # noqa: E501
triton_kernel_moe_forward,
)