MiniCPM
MiniCPM copied to clipboard
[Bug]: openai_api_server_demo.generate_minicpm:158 - Input length larger than 20
Is there an existing issue ? / 是否已有相关的 issue ?
- [X] I have searched, and there is no existing issue. / 我已经搜索过了,没有相关的 issue。
Describe the bug / 描述这个 bug
使用 demo/openai_api_demo/openai_api_server_demo.py 无法正常对话,model.config.max_length拿到的值只有20
To Reproduce / 如何复现
- 运行demo/openai_api_demo/openai_api_server_demo.py,等待服务启动
- 修改demo/openai_api_demo/openai_api_request_demo.py的第二十一行为:
"content": "我有1000元人民币,在超市花掉了80元,还剩多少?" - 然后就能在server端看到错误日志:
minicpm-1 | 2024-09-07 02:00:01.108 | DEBUG | __main__:create_chat_completion:240 - ==== request ====
minicpm-1 | {'messages': [ChatMessage(role='user', content='我有1000元人民币,在超市花掉了80元,还剩多少?', name=None)], 'temperature': 0.8, 'top_p': 0.8, 'max_tokens': 4096, 'echo': False, 'repetition_penalty': 1.1, 'tools': None}
minicpm-1 | 2024-09-07 02:00:01.109 | ERROR | __main__:generate_minicpm:158 - Input length larger than 20
minicpm-1 | 2024-09-07 02:00:01.109 | DEBUG | __main__:create_chat_completion:276 - ==== message ====
minicpm-1 | role='assistant' content='' name=None
minicpm-1 | INFO: 172.21.0.1:34988 - "POST /v1/chat/completions HTTP/1.1" 200 OK
Expected behavior / 期望的结果
能正常对话
Screenshots / 截图
No response
Environment / 环境
- OS: [Docker Ubuntu 22.04]
- Pytorch: [torch '2.4.1+cu121']
- CUDA: [CUDA 12.1.1]
- Device: [RTX3090]
Additional context / 其他信息
No response
你好,你可以修改model.config.max_length这个值,可以在config文件中直接修改,20并不是一个有效的值
辛苦明确下使用的是哪个MiniCPM模型?请参考这里的回复