Question: Any plan to support more chat products?
SherL0cked opened this issue · comments
Firstly, I want to thank you for your exceptional work on revChatGPT and ChatGPT-Proxy, your efforts are greatly appreciated.
Out of curiosity, I was wondering if there are plans to expand support to more chat platforms, especially considering the rise of several public chat products like Anthropic, Bard, Vicuna, they're becoming increasingly popular and seem to maintain a fairly similar and standard approach to hosting chat models.
It seems that the development of a unified SDK for experimenting with and supporting various models could significantly benefit developers by reducing cost and complexity. Projects like FastChat are doing similar things but with open-source models.
And as we're all aware there will inevitably be increased restrictions imposed by service providers in the future, and it would be great to have your thoughts on it.
Question: Any plan to support more chat products?
I'm already supporting a few others:
https://github.com/acheong08/Bard
https://github.com/acheong08/EdgeGPT
Vicuna is for running locally and you can find some pretty good bindings for llama.cpp or gpt4all
I have applied for access to Anthropic but have yet to get a response
I'm planning on unifying all my current API based projects into one to simplify usage. I will work on that over the coming weeks as my exams have just ended
And as we're all aware there will inevitably be increased restrictions imposed by service providers in the future, and it would be great to have your thoughts on it.
Restrictions are expected. For "uncensored" models, there are plenty available to run locally like gpt4-x-vicuna (there are probably more recent ones but I've been busy with exams and haven't kept up)
Awesome, integration of different products into one sounds really exciting! There's one repo ChatAll is moving in similar direction but with less flexibility and programmability ... For open-source models, there's actually an online version in https://chat.lmsys.org/ which might able to be leveraged?
I've been busy as hell. Will try to work on this soon.
https://github.com/acheong08/ChatGPT-to-API has been refactored with this in mind. Working with websockets for Bing is a nightmare. Bard will probably be the first to be supported