LibreChat – Enhanced ChatGPT Clone(github.com) |
LibreChat – Enhanced ChatGPT Clone(github.com) |
If you're interested in a fully Libre LLM stack, I've had fun lately with ollama [0] and ollama-webui [1]. It was pretty trivial to take ollama-webui's docker-compose file and set up a locally-running chat server with Mistral 7B. Trying out different models and prompts was likewise very easy to get started with.
Mistral isn't anything like as good as GPT-4, but it's Apache licensed and fully local, which meets my definition of Libre. I'll continue to use both while the FOSS stacks catch up, but it's fun to keep up with the progress on the open source stuff as tooling develops.
Related issue: https://github.com/danny-avila/LibreChat/issues/1131
LiteLLM: https://github.com/BerriAI/litellm
Guide: https://github.com/danny-avila/LibreChat/blob/main/docs/inst...
I still find it a bit concerning that even the LibreChat docs about LiteLLM don't mention local models in the list of use cases. It looks like there's a GitHub ticket to add that to the docs [0], which would be good.
Maybe I'm just growing old and crotchety, but it makes me uncomfortable to see a word like Libre—which was specifically coined in English to differentiate itself from Free-as-in-Beer—used in a context where monetary cost seems to be the primary concern in the maintainers' minds.
(From the naming, I was expecting some "strongly free software", like LibreJS, Libreboot, linux-libre, which would not rely on something closed like ChatGPT - I guess if someone made a free LLM-based chat, they could call it ChatGPL (though they should probably not, too close to ChatGPT))
If I'm ever stuck in a situation where I need to use ChatGPT, I'll consider using LibreChat :-)
What does this have that kobold.cpp, llama.cpp backends + sillytavern et al frontends with one of the thousand llm weights won't give you?
For entertainment, these offer simultaneous llm output with methods to retain context, allows outputs that can be vocalized via TTS voice simulation, inputs via mucrophone, and can provide illustrations of content via stable diffusion, and allow multiple chat bot "characters" to be in the same conversation, all of which honestly gets a bit surreal.
I'm not being critical - just curious.
What you actually are looking for is gpt4all:
On the other hand, they dont train with API usage. So something like this is very interesting.
Question: does LibreChat support importing OpenAI history?
so some privacy and multimodal is great. glad the pieces are getting more user friendly to casually use