Add a free gpt3-4 provider to avoid having to use ChatGPT.

[GhomKrosmonaute]

Add a free gpt3-4 provider to avoid having to use ChatGPT.

Initial proposal

Hello !

Currently, OpenCommit uses ChatGPT as its AI model for generating commit messages. However, using a cloud-based AI model like ChatGPT raises privacy concerns for some users who do not want their private project data to be stored on external servers. Therefore, I propose adding a free, open-source GPT-3 or GPT-4 provider, such as llama.cpp, to OpenCommit as an alternative to using ChatGPT. This would allow users to use local language models to generate their commit messages, without having to worry about the privacy of their data.

Edit 03-30-2023

Prevent breaking changes

• Add a LLAMA_LOCATION optional config entry for llama.cpp/main path.
• Add a provider config entry with "openai" as default value (can be "llama").

[di-sukharev]

great idea man, do you want to make a PR?

[GhomKrosmonaute]

Thx, great tool! 😁

I think it would take too long to make a PR, I'm overwhelmed with projects... But I'm willing to create technical documentation about it, if it helps. However, it is still possible that I create a gpt provider module by coding sometimes!

[GhomKrosmonaute]

Well, I started to implement this via a fork of dev branch here.

I just add compatibility with llama.cpp, it will certainly show the way to those who would like to make opencommit compatible with their own language model 👍

[di-sukharev]

@GhomKrosmonaute would be awesome to have the tech doc from you <3

[Ed1123]

I can also work on adding another language model so it also works offline. That documentation will be awesome to have, @GhomKrosmonaute.

[github-actions[bot]]

Stale issue message

[di-sukharev]

we now support Ollama, anyway reopening @GhomKrosmonaute @Ed1123 is you guys want to work on this