Make fastchat.serve.model_worker to take debug argument (#2628)

Co-authored-by: hi-jin <[email protected]>

fastchat/serve/model_worker.py

@@ -58,6 +58,7 @@ def __init__(
         conv_template: Optional[str] = None,
         embed_in_truncate: bool = False,
         seed: Optional[int] = None,
+        debug: bool = False,
         **kwargs,
     ):
         super().__init__(
@@ -82,6 +83,7 @@ def __init__(
             gptq_config=gptq_config,
             awq_config=awq_config,
             exllama_config=exllama_config,
+            debug=debug,
         )
         self.device = device
         if self.tokenizer.pad_token == None:
@@ -283,6 +285,9 @@ def create_model_worker():
         default=None,
         help="Overwrite the random seed for each generation.",
     )
+    parser.add_argument(
+        "--debug", type=bool, default=False, help="Print debugging messages"
+    )
     args = parser.parse_args()
     logger.info(f"args: {args}")
 
@@ -333,6 +338,7 @@ def create_model_worker():
         conv_template=args.conv_template,
         embed_in_truncate=args.embed_in_truncate,
         seed=args.seed,
+        debug=args.debug,
     )
     return args, worker