Update chat.py

chat.py

@@ -48,7 +48,7 @@ def generate_response(message, history):
         response = tokenizer.decode(output[0], skip_special_tokens=True)
         
         # Extract only the assistant's response
-        assistant_response = response.rsplit("<|im_start|> assistant\n", 1)[1]
+        assistant_response = response.split("<|im_start|>assistant\n")[-1]
         
         return assistant_response
     except Exception as e: