Support redundant experts in expert parallel (#6461)

python/sglang/srt/managers/expert_location.py

@@ -163,8 +163,7 @@ class ExpertLocationMetadata:
 
         num_physical_experts = (
             model_config_for_expert_location.num_logical_experts
-            # TODO pr-chain: enable this later
-            # + server_args.ep_num_redundant_experts
+            + server_args.ep_num_redundant_experts
         )
         ep_size = server_args.ep_size
         assert num_physical_experts % ep_size == 0

python/sglang/srt/managers/schedule_batch.py

@@ -90,6 +90,7 @@ global_server_args_dict = {
     "speculative_accept_threshold_single": ServerArgs.speculative_accept_threshold_single,
     "torchao_config": ServerArgs.torchao_config,
     "triton_attention_reduce_in_fp32": ServerArgs.triton_attention_reduce_in_fp32,
+    "ep_num_redundant_experts": ServerArgs.ep_num_redundant_experts,
 }
 
 logger = logging.getLogger(__name__)

python/sglang/srt/model_executor/model_runner.py

@@ -206,6 +206,7 @@ class ModelRunner:
                 "speculative_accept_threshold_acc": server_args.speculative_accept_threshold_acc,
                 "use_mla_backend": self.use_mla_backend,
                 "mm_attention_backend": server_args.mm_attention_backend,
+                "ep_num_redundant_experts": server_args.ep_num_redundant_experts,
             }
         )
 

python/sglang/srt/models/deepseek_v2.py

@@ -243,7 +243,9 @@ class DeepseekV2MoE(nn.Module):
         self.gate = MoEGate(config=config, prefix=add_prefix("gate", prefix))
 
         self.experts = get_moe_impl_class()(
-            num_experts=config.n_routed_experts + self.n_share_experts_fusion,
+            num_experts=config.n_routed_experts
+            + self.n_share_experts_fusion
+            + global_server_args_dict["ep_num_redundant_experts"],
             top_k=config.num_experts_per_tok + min(self.n_share_experts_fusion, 1),
             hidden_size=config.hidden_size,
             intermediate_size=config.moe_intermediate_size,
@@ -285,7 +287,10 @@ class DeepseekV2MoE(nn.Module):
         if global_server_args_dict["enable_deepep_moe"]:
             # TODO: we will support tp < ep in the future
             self.ep_size = get_tensor_model_parallel_world_size()
-            self.num_experts = config.n_routed_experts
+            self.num_experts = (
+                config.n_routed_experts
+                + global_server_args_dict["ep_num_redundant_experts"]
+            )
             self.renormalize = config.norm_topk_prob
             self.topk_group = config.topk_group
             self.num_expert_group = config.n_group
@@ -299,7 +304,7 @@ class DeepseekV2MoE(nn.Module):
                 group=parallel_state.get_tp_group().device_group,
                 router_topk=self.top_k,
                 permute_fusion=True,
-                num_experts=config.n_routed_experts,
+                num_experts=self.num_experts,
                 num_local_experts=config.n_routed_experts // self.tp_size,
                 hidden_size=config.hidden_size,
                 params_dtype=config.torch_dtype,

python/sglang/srt/server_args.py

@@ -170,6 +170,7 @@ class ServerArgs:
     enable_ep_moe: bool = False
     enable_deepep_moe: bool = False
     deepep_mode: Optional[Literal["auto", "normal", "low_latency"]] = "auto"
+    ep_num_redundant_experts: int = 0
     ep_dispatch_algorithm: Optional[Literal["static", "dynamic"]] = None
     init_expert_location: str = "trivial"
     expert_distribution_recorder_mode: Optional[
@@ -1273,6 +1274,12 @@ class ServerArgs:
             default="auto",
             help="Select the mode when enable DeepEP MoE, could be `normal`, `low_latency` or `auto`. Default is `auto`, which means `low_latency` for decode batch and `normal` for prefill batch.",
         )
+        parser.add_argument(
+            "--ep-num-redundant-experts",
+            type=int,
+            default=ServerArgs.ep_num_redundant_experts,
+            help="Allocate this number of redundant experts in expert parallel.",
+        )
         parser.add_argument(
             "--ep-dispatch-algorithm",
             type=str,