provide some overviews of models / gpus to run these as a "tricks and tips"?

[jerabaul29]

Thanks for an amazing project!

One of the challenges when getting started with these llm lsps is getting a feeling for what is reasonable to run on different hardware.

Would it be within scope to have a dedicated documentation section with a few tables, recommending what model / config to use for different hardware? For example if I have on a CPU, what is a reasonable LLM to run? If I have a 4080 GPU, what is a good choice? I wonder if it feels a bit like a jungle otherwise for newcomers (at least it is my feeling now :) ).

[SilasMarvin]

This is a really good point. This project has an education problem. To use this project effectively you must not only have a decent understanding of LLMs, you must have a good understanding of the LSP.

This is something I have been thinking about and I have a few ideas. The main one being an auto configurator website where you are asked a series of questions and tick a series of boxes and are given a configuration.

Stay tuned.