Hacker News Reader

You can use it to directly search any models, download and run 100% locally.

M-series Macs are the simplest, they Just Work. Even faster if you tick the GPU box.

Windows needs the right kind of GPU to get workable speed.

Can that interact with files on your computer, like they show in the video?

Last time I used it, LM Studio doesn't include RAG.

I just installed it yesterday and you are right it does not seem to have RAG but you can use something like anythingLLM to do the rag work and ut has built in integration with studio LM.

You could try my Mistral implementation: https://github.com/Const-me/Cgml/blob/master/Mistral/Mistral...

You can start with Kobold.cpp which should handhold you through the process.

Trying it now. Doesn't seem censored to me.

it uses Mistral or Llama 2