I have attempted to install AlphaCodium and configured the LLM's model and key according to the instructions. Does this mean I can now connect to the LLM? Is there no need to configure a base_url?

[zte-yangjizhe]

I have attempted to install AlphaCodium and configured the LLM's model and key according to the instructions. Does this mean I can now connect to the LLM? Is there no need to configure a base_url?

[mrT23]

what do you mean by "base_rul" ??
you are working with openai models. Their interface is defined here:
https://github.com/Codium-ai/AlphaCodium/blob/main/alpha_codium/llm/ai_handler.py#L108

[zte-yangjizhe]

Oh, sorry, what I meant was api_base.
I have two questions: First, does AlphaCodium support all of OpenAI's LLMs? I used an internally developed model called CodeLlama's embedded model, but it seems to fail when running.

Second, what is the minimum Python version requirement? I noticed that the Dockerfile specifies version 3.9.

[zte-yangjizhe]

I tried installing Python 3.9.1, but encountered the following error during execution

[zte-yangjizhe]

and more：

[zte-yangjizhe]

[mrT23]

"AlphaCodium" (i.e litellm) supports any openai model from:
https://platform.openai.com/docs/models/gpt-4-and-gpt-4-turbo
https://platform.openai.com/docs/models/gpt-3-5

Regarding other models, you need to debug and edit if needed whats happening in:
https://github.com/Codium-ai/AlphaCodium/blob/main/alpha_codium/llm/ai_handler.py#L108

[zte-yangjizhe]

Can you provide me with a list of the successful version information you have tested? For example, versions of OpenAI, litellm, OpenSSL, etc.

[mrT23]

No.
And thats not the problem you experience.
Try first with a standard openai model, and make sure it works. it would.
than debug your problem with your custom model

all the best

[zte-yangjizhe]

Thank you, I have successfully executed it