run on cpu

app.py

@@ -444,15 +444,16 @@ if __name__ == "__main__":
     logger = logging.getLogger(__name__)
     logger.info(gr.__version__)
     args = parse_args()
-    model_name = args.model_name
+    model_name = "tinyllava_demo"
     tokenizer, model, image_processor, context_len = load_pretrained_model(
-        model_path=args.model_path,
-        model_base=None,
-        model_name=args.model_name,
-        load_4bit=args.load_4bit,
-        load_8bit=args.load_8bit
+        model_path=model_name,
+        model_base="bczhou/TinyLLaVA-3.1B",
+        model_name=get_model_name_from_path(model_name + "_lora"),
+        device="cpu",
+        load_4bit=False,
+        load_8bit=False
     )
 
     demo = build_demo()
     demo.queue()
-    demo.launch(server_name=args.host, server_port=args.port, share=args.share)
+    demo.launch()