A discord bot using the research llama model
I was able to run the LLaMA models on 3080Ti. (research only) As it's a quick hack, the code will get cleaned and uploaded soon. Meanwhile, if you want to try it, I hosted a discord bot (running the 13B version) for my server (free access).
The server invite is https://discord.gg/SgmBydQ2Mn.
There is also a video demo: https://youtu.be/CRI0RMrgCuo.