How do we finetune the model with new data?

[ekolawole]

Can we have a finetune.cpp or finetune.exe file to incorporate new data into the model? The use case will be to design an AI model that can do more than just general chat. It can become very knowledgeable in specific topics they are finetuned on. Also, after creating the finetune.exe , please ensure no GPU is required for the entire process. Because that is what makes this repo awesome in the first place.

[Terristen]

From my limited research, the recommendations are to rent A100 time in the cloud and run your finetuning there, like Stanford did. There's a project that's trying to clean up the Alpaca training data, and from what I can tell, that will fix a lot of the issues we're seeing in 7B and maybe 13B (once the model is retrained on that new cleaner data).

There's just no good way to train these models via CPU. What an A100 can do in 1 hour would take my 12-core 24-thread CPU a week or more.