[MoE Refactor] Rename FusedMoE.make_expert_params_mapping to fused_moe_make_expert_params_mapping (#40671)

Signed-off-by: Bill Nell <bnell@redhat.com>
This commit is contained in:
bnellnm
2026-04-23 11:22:34 -04:00
committed by GitHub
parent 8824f50f1f
commit 1c2c1eb8b9
53 changed files with 254 additions and 98 deletions
+6 -4
View File
@@ -41,7 +41,9 @@ from vllm.distributed import (
get_pp_group,
)
from vllm.logger import init_logger
from vllm.model_executor.layers.fused_moe import FusedMoE
from vllm.model_executor.layers.fused_moe import (
fused_moe_make_expert_params_mapping,
)
from vllm.model_executor.layers.layernorm import RMSNorm
from vllm.model_executor.layers.logits_processor import LogitsProcessor
from vllm.model_executor.layers.vocab_parallel_embedding import (
@@ -308,7 +310,7 @@ class Glm4MoeLiteModel(nn.Module):
def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
# Params for weights, fp8 weight scales, fp8 activation scales
# (param_name, weight_name, expert_id, shard_id)
return FusedMoE.make_expert_params_mapping(
return fused_moe_make_expert_params_mapping(
self,
ckpt_gate_proj_name="gate_proj",
ckpt_down_proj_name="down_proj",
@@ -334,7 +336,7 @@ class Glm4MoeLiteModel(nn.Module):
# Params for weights, fp8 weight scales, fp8 activation scales
# (param_name, weight_name, expert_id, shard_id)
expert_params_mapping = FusedMoE.make_expert_params_mapping(
expert_params_mapping = fused_moe_make_expert_params_mapping(
self,
ckpt_gate_proj_name="gate_proj",
ckpt_down_proj_name="down_proj",
@@ -616,7 +618,7 @@ class Glm4MoeLiteForCausalLM(
def get_expert_mapping(self) -> list[tuple[str, str, int, str]]:
# Params for weights, fp8 weight scales, fp8 activation scales
# (param_name, weight_name, expert_id, shard_id)
return FusedMoE.make_expert_params_mapping(
return fused_moe_make_expert_params_mapping(
self,
ckpt_gate_proj_name="gate_proj",
ckpt_down_proj_name="down_proj",