Update app.py

app.py

@@ -50,7 +50,7 @@ def load_model_norm():
     # For example: revision="main"
     model = AutoModelForCausalLM.from_pretrained(model_name_or_path,device_map="auto", trust_remote_code=True,revision="gptq-4bit-128g-actorder_True")
     # Switch to CPU inference
-    model.cpu()
+    model.to("cuda")
     tokenizer = AutoTokenizer.from_pretrained(model_name_or_path, use_fast=True)
 
     return model, tokenizer