[ROCm][Bugfix] Re-tag AITER MoE weights as preshuffled after replace_parameter (#42061)

Signed-off-by: Markus Hartikainen <markus.hartikainen@amd.com>
Co-authored-by: TJian <tunjian.tan@embeddedllm.com>
This commit is contained in:
Markus Hartikainen
2026-05-08 21:42:07 -07:00
committed by GitHub
co-authored by TJian
parent df2636a9d8
commit e8f9038ebd
3 changed files with 21 additions and 0 deletions
@@ -28,6 +28,7 @@ from vllm.model_executor.layers.fused_moe.config import (
)
from vllm.model_executor.layers.fused_moe.layer import UnquantizedFusedMoEMethod
from vllm.model_executor.layers.fused_moe.oracle.fp8 import (
Fp8MoeBackend,
convert_to_fp8_moe_kernel_format,
make_fp8_moe_kernel,
make_fp8_moe_quant_config,
@@ -766,6 +767,11 @@ class Fp8MoEMethod(FusedMoEMethodBase):
replace_parameter(layer, f"w13_{self.weight_scale_name}", w13_scale)
replace_parameter(layer, f"w2_{self.weight_scale_name}", w2_scale)
# AITER backend requires weights to be marked as shuffled.
if self.fp8_backend == Fp8MoeBackend.AITER:
layer.w13_weight.is_shuffled = True
layer.w2_weight.is_shuffled = True
self.moe_quant_config = self.get_fused_moe_quant_config(layer)
if self.moe_quant_config:
assert self.experts_cls is not None