clamuu

joined 10 months ago

[P] X—LLM: Few lines of code to train your own 7B LLM in Colab using cutting edge techniques like QLoRA in c/machinelearning@academy.garden

[–] clamuu@alien.top 1 points 10 months ago

Wow thank you for the detailed reply. Your library looks fantastic. I'm definitely going to give it a go. I'm going to try fine-tuning it on music theory. Is that a crazy idea? Training on a GPU sounds much better. I looked more thoroughly through the repo and found it's all explained in there.

Telegram is a popular app here in the UK. Seems to me like an excellent way to launch it as there's no need for the user to download an app. WhatsApp is much more popular here but maybe it's harder to deploy a bot like this on WhatsApp?

[P] X—LLM: Few lines of code to train your own 7B LLM in Colab using cutting edge techniques like QLoRA in c/machinelearning@academy.garden

[–] clamuu@alien.top 1 points 10 months ago (2 children)

Is this something that could be trained on a laptop without a gpu or would it be better to use cloud based GPU services?

Also can you or anyone else recommend any other libraries which simplify LLM training? I've done some ML projects but I'd like to do something a bit deeper and this looks perfect.

I tried out Talequest by the way. Very impressive.