chenhaoqiang

调试了下代码，发现是这里的代码问题，没有从JmReportTokenServiceI.getUserInfo()获取用户信息传入   

> Sorry for the delayed answer. > > Logical structures such as if/else are not currently supported in the sequence diagram. Since the currently used library javassist for analyzing the...

> Bytecode is currently used for sequence analysis, as this offers greater flexibility, even across library boundaries. This would be different with JavaParser as far as I can see at...

 自己找到了，第一个版本暂未实现。

> 第一张图片应该是因为使用了ocr的问题。我这边也有出现这情况。 > > 本地部署，感觉效果不如开放的demo那么好。相同的图片和问题，本地回答效果不怎么好，而且还会重复回答直到最大长度。 可能是7b的原因，34b跑不动，没算力了😂

> Hi, thanks for your report! The 7B model is actually far from the implemented 34B demo. Please try 34B model for a better experience. And we plan to support...

 但是postman有返回的，只是调用一次后服务就挂了

> 不好意思，我的环境复现不出来。从报错信息看，猜测可能和某些库的版本或者路径有关。 > > 可能有用的信息： > > [https://github.com/tensorflow/tensorflow/issues/6968](https://github.com/tensorflow/tensorflow/issues/6968) 改成这样启动就可以了！谢谢回复！ gunicorn -b 0.0.0.0:5000 --workers=1 my_openai_api:app

> modify protocol/openai_api_protocol.py > > ```python > class ChatCompletionRequest(BaseModel): > model: str > messages: Union[str, List[Dict[str, str]]] > temperature: Optional[float] = 0.7 > top_p: Optional[float] = 1.0 > n: Optional[int]...