[update]

main.py

@@ -159,7 +159,7 @@ def chat_with_llm_streaming(question: str,
 
     for encoded_utterance in encoded_utterances:
         input_ids.extend(encoded_utterance)
-        if model.config.model_type == "chatglm":
+        if model.config.model_type != "chatglm":
             input_ids.append(tokenizer.eos_token_id)
 
     input_ids = torch.tensor([input_ids], dtype=torch.long)