Could we have a written walkthrough of finetuning llama/mistral with this?

[azambuilds]

I'm really excited with finetuning LLMs on rust but I'm a complete beginner to machine learning so even though I've spent some time going through the repo it would be great if there was a plainwritten tutorial or step by step guide on finetuning some of the more popular models with this.

Thanks!

[EricLBuehler]

@Tameflame, yes, absolutely. I'll take a look at adding this very soon.

[jamesfebin]

Any update on this?

[AlpineVibrations]

this would be super great. is there any updates?

[azambuilds]

checking in on this issue, would really appreciate any guidance on this!

[vrama628]

Hi all, it took a fair bit of troubleshooting and required modifications to both candle-lora and candle-core itself, but I've finally successfully used this crate to fine-tune Llama 3 8B. @EricLBuehler if I put together a guide on this, where would be the best place for me to put it? This crate doesn't seem to have a book like candle itself does -- it might make sense for me to contribute it to the main candle repo under candle-book, but it also seems that publishing that manual has been temporarily excluded from CD.

[EricLBuehler]

Hi @vrama628!

Thanks for taking the time to do that. I think the best place for a guide about fine-tuning would be a HF hub blogpost. If you could write that, then we can link it on this crate as well as candle itself.

Does this sound good?