[MoE Refactor] Remove SharedFusedMoE class (#35782)
Signed-off-by: Bill Nell <bnell@redhat.com>
This commit is contained in:
@@ -44,7 +44,7 @@ from vllm.model_executor.layers.attention import (
|
||||
Attention,
|
||||
StaticSinkAttention,
|
||||
)
|
||||
from vllm.model_executor.layers.fused_moe import SharedFusedMoE
|
||||
from vllm.model_executor.layers.fused_moe import FusedMoE
|
||||
from vllm.model_executor.layers.layernorm import RMSNorm
|
||||
from vllm.model_executor.layers.linear import (
|
||||
ColumnParallelLinear,
|
||||
@@ -200,7 +200,7 @@ class OpenPanguMoE(nn.Module):
|
||||
else:
|
||||
self.shared_experts = None
|
||||
|
||||
self.experts = SharedFusedMoE(
|
||||
self.experts = FusedMoE(
|
||||
shared_experts=self.shared_experts,
|
||||
num_experts=config.n_routed_experts,
|
||||
top_k=config.num_experts_per_tok,
|
||||
@@ -1149,7 +1149,7 @@ class OpenPanguModel(nn.Module):
|
||||
]
|
||||
has_experts = hasattr(self.config, "n_routed_experts")
|
||||
if has_experts:
|
||||
expert_merge_mapping = SharedFusedMoE.make_expert_params_mapping(
|
||||
expert_merge_mapping = FusedMoE.make_expert_params_mapping(
|
||||
self,
|
||||
ckpt_gate_proj_name="gate_proj",
|
||||
ckpt_down_proj_name="down_proj",
|
||||
|
||||
Reference in New Issue
Block a user