[Refactor] Delete dead code in models (#51838)

Signed-off-by: yewentao256 <zhyanwentao@126.com>
This commit is contained in:
Wentao Ye
2026-08-11 15:44:26 -04:00
committed by GitHub
parent ded6c452c3
commit fd04edef3e
25 changed files with 34 additions and 223 deletions
+3 -15
View File
@@ -293,7 +293,6 @@ class DeepseekV2MoE(nn.Module):
self.routed_scaling_factor = getattr(config, "routed_scaling_factor", 1.0)
self.ep_group = get_ep_group().device_group
self.ep_rank = get_ep_group().rank_in_group
self.ep_size = self.ep_group.size()
self.n_routed_experts: int = config.n_routed_experts
self.n_shared_experts: int = config.n_shared_experts
@@ -329,11 +328,6 @@ class DeepseekV2MoE(nn.Module):
self.n_physical_experts = self.n_logical_experts + self.n_redundant_experts
self.n_local_physical_experts = self.n_physical_experts // self.ep_size
self.physical_expert_start = self.ep_rank * self.n_local_physical_experts
self.physical_expert_end = (
self.physical_expert_start + self.n_local_physical_experts
)
self.is_rocm_aiter_moe_enabled = rocm_aiter_ops.is_fused_moe_enabled()
self.is_fusion_moe_shared_experts_enabled = (
rocm_aiter_ops.is_fusion_moe_shared_experts_enabled()
@@ -409,15 +403,9 @@ class DeepseekV2MoE(nn.Module):
if self.is_sequence_parallel and not already_sequence_parallel:
hidden_states = sequence_parallel_chunk(hidden_states)
if self.experts.is_internal_router:
final_hidden_states = self.experts(
hidden_states=hidden_states, router_logits=hidden_states
)
else:
router_logits, _ = self.gate(hidden_states)
final_hidden_states = self.experts(
hidden_states=hidden_states, router_logits=router_logits
)
final_hidden_states = self.experts(
hidden_states=hidden_states, router_logits=hidden_states
)
if self.is_sequence_parallel and not already_sequence_parallel:
final_hidden_states = tensor_model_parallel_all_gather(