Update chat.py

chat.py

@@ -48,11 +48,7 @@ def generate_response(message, history):
         response = tokenizer.decode(output[0], skip_special_tokens=True)
         
         # Extract only the assistant's response
-        if "<|im_start|>assistant\n" in response and "<|im_end|>" in response:
-            assistant_response = response.split("<|im_start|>assistant\n")[1].split("<|im_end|>")[0]
-        else:
-            # Fallback: Return the entire generated response if special tokens are not found
-            assistant_response = response
+        assistant_response = response.split("<|im_start|>assistant\n")[1].split("<|im_end|>")[0]
         
         return assistant_response
     except Exception as e: