Update app.py

app.py

@@ -10,12 +10,11 @@ model = AutoModelForCausalLM.from_pretrained(checkpoint).to(device)
 @spaces.GPU(duration=120)
 def predict(message, history):
     history.append({"role": "user", "content": message})
-    input_text = tokenizer.apply_chat_template(history, tokenize=False)
-    print(input_text)
+    input_text = tokenizer.apply_chat_template(history, tokenize=False) + "<|start_header_id|>assistant<|end_header_id|>\n\n"
     inputs = tokenizer.encode(input_text, return_tensors="pt").to(device)  
     outputs = model.generate(inputs, max_new_tokens=100, temperature=0.7, top_p=0.9, do_sample=True)
     decoded = tokenizer.decode(outputs[0])
-    response = decoded
+    response = decoded.split("<|start_header_id|>assistant<|end_header_id|>\n\n")[-1]
     return response
 
 demo = gr.ChatInterface(predict, type="messages")