[BUG/Help] <title>6b-chat基模型输出bug

[HaomingX]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

基模型推理，任何问题都会在回答问题之后输出训练数据

温度低时输出完数据还会一直重复某些句子

Expected Behavior

No response

Steps To Reproduce

1. 按照huggingface上的配置
2. 导入和参数设置
   tokenizer = AutoTokenizer.from_pretrained(model_path, trust_remote_code=True)
   model = AutoModelForCausalLM.from_pretrained(model_path, trust_remote_code=True, device='cuda')

response, _history = model.chat(tokenizer, request, history=history, max_length= 512, num_beams=1,
do_sample=True, top_p=0.8, temperature=0.1)

Environment

- OS:Ubuntu 20.04
- Python:3.10
- Transformers:4.33.2
- PyTorch:
- CUDA Support (`python -c "import torch; print(torch.cuda.is_available())"`) :True

Anything else?

No response