Create clear instructions for downloading and converting the models

[edmundronald]

Clear instructions are needed to allow new arrivals to download and convert the models, in spite of the multiple format versions (non quantised, quantised, various llama versions etc) .

I would suggest that each llama or alpaca etc print a version on startup, and that the conversions scripts have this in their name, and also that a program reading a file and figuring out what it is from a magic print the version read and the version expected even if it aborts.

Edmund

[gjmulder]

There are extremely clear instructions on obtaining the LLaMA models at https://github.com/ggerganov/llama.cpp#obtaining-and-verifying-the-facebook-llama-original-model-and-stanford-alpaca-model-data

[sw]

Hello thumbs-down brigade!

Closing this as the README is quite clear in how to use llama.cpp. We don't distribute Facebook's model data, but there are alternatives. You may want to check out OpenLLaMA, though that doesn't quite seem to work with llama.cpp -> #1291