Update app.py

app.py

@@ -11,11 +11,17 @@ model_id = "TheBloke/Yi-34B-200K-Llamafied-GPTQ"
 
 gptq_config = GPTQConfig( bits=4, exllama_config={"version": 2})
 tokenizer = YiTokenizer.from_pretrained("./")
-model = AutoModelForCausalLM.from_pretrained( model_id, device_map="cuda", quantization_config=gptq_config)
+model = AutoModelForCausalLM.from_pretrained( model_id, device_map="auto", torch_dtype="auto", trust_remote_code=True, quantization_config=gptq_config)
 
 def run(message, chat_history, max_new_tokens=4056, temperature=3.5, top_p=0.9, top_k=800):
     prompt = get_prompt(message, chat_history)
     input_ids = tokenizer.encode(prompt, return_tensors='pt')
+    
+    print("Input IDs:", input_ids)  # Debug print
+    print("Input IDs shape:", input_ids.shape)  # Debug print
+    if input_ids.shape[1] == 0:
+        raise ValueError("The input is empty after tokenization.")
+
     input_ids = input_ids.to(model.device)
     response_ids = model.generate(
         input_ids,