There is an Exception when i deploy this project and ask questions.

[liys6666]

[jinliyl]

Is there a dashscope key injected into env, or this dashscope key can be used to call the Baichuan4 model directly?

[liys6666]

Yes, i have one. But it seems not support Baichuan4. If i want to use other models except OPENAI and QWEN，what should i do?

[jinliyl]

You need to implement a client for calling Baichuan4 in memoryscope.core.models, maintaining the same interface as the base class.

[liys6666]

Does it mean that I need to create a Baichuan4 Client `.py` file similar to the `base_model.py` shown in the picture? However, I am unable to find the location of the `api_url` corresponding to different models.

[jinliyl]

Yes, you can write a client for Baichuan4 based on the Basemodel framework. You can refer to the LlamaIndexGenerationModel for guidance. If you are familiar with llamaindex, a simpler way is to add a class corresponding to Baichuan in the llama_index at line 27 of the LlamaIndexGenerationModel.

[liys6666]

I am not familiar with llamaindex. I have to learn it now. So it means that we don't need a api_url, just a Baichuan4 llamaindex is enough.

[jinliyl]

Yes, accessing through llamaindex is the most convenient.

[liys6666]

Thank you, now i will have a try.