Update app.py

app.py

@@ -24,9 +24,9 @@ if not torch.cuda.is_available():
 
 if torch.cuda.is_available():
     model_id = "hon9kon9ize/CantoneseLLMChat-v1.0-7B"
-    model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto")
+    model = AutoModelForCausalLM.from_pretrained(model_id, device_map="auto", torch_dtype=torch.bfloat16, trust_remote_code=True)
     model = torch.compile(model)
-    tokenizer = AutoTokenizer.from_pretrained(model_id)
+    tokenizer = AutoTokenizer.from_pretrained(model_id, trust_remote_code=True)
     tokenizer.use_default_system_prompt = False