Update app.py

app.py

@@ -712,7 +712,8 @@ if __name__ == "__main__":
     )
     # argparser.add_argument("--model-path", type=str, default="facebook/opt-350m")
     argparser.add_argument("--model-base", type=str, default=None)
-    argparser.add_argument("--num-gpus", type=int, default=1)
+    argparser.add_argument("--device_map", type=str, default="cpu")
+    # argparser.add_argument("--num-gpus", type=int, default=1)
     argparser.add_argument("--conv-mode", type=str, default=None)
     argparser.add_argument("--temperature", type=float, default=0.7)
     argparser.add_argument("--max-new-tokens", type=int, default=4096)
@@ -729,6 +730,6 @@ if __name__ == "__main__":
     #model_name = get_model_name_from_path(args.model_path)
     model_name = "pangea_llava_qwen"
     tokenizer, model, image_processor, context_len = load_pretrained_model(model_path, None, model_name, args.load_8bit, args.load_4bit, **args)
-    model=model.to(torch.device('cuda'))
+    #model=model.to(torch.device('cuda'))
     chat_image_num = 0
     demo.launch()