Remove --modelopt-config in server_args (#2758)

python/sglang/srt/model_executor/model_runner.py

@@ -150,7 +150,6 @@ class ModelRunner:
                 "enable_nan_detection": server_args.enable_nan_detection,
                 "enable_dp_attention": server_args.enable_dp_attention,
                 "enable_ep_moe": server_args.enable_ep_moe,
-                "modelopt_config": server_args.modelopt_config,
             }
         )
 

python/sglang/srt/server_args.py

@@ -149,7 +149,6 @@ class ServerArgs:
     torch_compile_max_bs: int = 32
     cuda_graph_max_bs: Optional[int] = None
     torchao_config: str = ""
-    modelopt_config: str = ""
     enable_nan_detection: bool = False
     enable_p2p_check: bool = False
     triton_attention_reduce_in_fp32: bool = False
@@ -810,12 +809,6 @@ class ServerArgs:
             default=ServerArgs.torchao_config,
             help="Optimize the model with torchao. Experimental feature. Current choices are: int8dq, int8wo, int4wo-<group_size>, fp8wo, fp8dq-per_tensor, fp8dq-per_row",
         )
-        parser.add_argument(
-            "--modelopt-config",
-            type=str,
-            default=ServerArgs.modelopt_config,
-            help="Optimize the model with nvidia-modelopt. Experimental feature. Current choices are: fp8",
-        )
         parser.add_argument(
             "--enable-nan-detection",
             action="store_true",