[MoE] Move cutlass moe to fused_moe/experts/ (#40574)
Signed-off-by: Jackmin801 <ongjackm@gmail.com> Co-authored-by: Claude <noreply@anthropic.com>
This commit is contained in:
+2
-2
@@ -14,7 +14,7 @@ from vllm.model_executor.layers.fused_moe.config import (
|
||||
FusedMoEQuantConfig,
|
||||
mxfp4_moe_quant_config,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe.cutlass_moe import (
|
||||
from vllm.model_executor.layers.fused_moe.experts.cutlass_moe import (
|
||||
CutlassExpertsMxfp4,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe.fused_marlin_moe import (
|
||||
@@ -149,7 +149,7 @@ class CompressedTensorsW4A4Mxfp4MoEMethod(CompressedTensorsMoEMethod):
|
||||
if self.use_cutlass_mxfp4:
|
||||
# Swizzle weight scales from flat checkpoint layout [E, N, K//32]
|
||||
# to CUTLASS tiled layout [E, numMTiles*numKTiles*512].
|
||||
from vllm.model_executor.layers.fused_moe.cutlass_moe import (
|
||||
from vllm.model_executor.layers.fused_moe.experts.cutlass_moe import (
|
||||
swizzle_mxfp4_scales,
|
||||
)
|
||||
|
||||
|
||||
+1
-1
@@ -315,7 +315,7 @@ class CompressedTensorsW4A8Fp8MoEMethod(CompressedTensorsMoEMethod):
|
||||
)
|
||||
assert self.moe_quant_config is not None
|
||||
|
||||
from vllm.model_executor.layers.fused_moe.cutlass_moe import (
|
||||
from vllm.model_executor.layers.fused_moe.experts.cutlass_moe import (
|
||||
cutlass_moe_w4a8_fp8,
|
||||
)
|
||||
|
||||
|
||||
Reference in New Issue
Block a user