Support for OpenELM - Githubissues

unslothai / unsloth

Finetune Llama 3.2, Mistral, Phi, Qwen & Gemma LLMs 2-5x faster with 80% less memory

https://unsloth.ai

Apache License 2.0

17.99k stars 1.25k forks source link

Open alvations opened 5 months ago

alvations commented 5 months ago

Would it be possible to support apple/OpenELM-3B-Instruct and apple/OpenELM-3B like how Phi-3 models are supported on the "Finetune for Free" section of the README

If users want to contribute to the repo, could you help to give some pointers on a couple of related questions:

Are there specific conversions that needs to be applied or ran to convert models to run with unsloth?
Is there some guide as to contributing and creating the similar colabs that you've created on the README?

danielhanchen commented 5 months ago

Yes we'll have automatic all model support in a future release, so it'll be supported!

danielhanchen commented 5 months ago

Oh normally I would take the Colab, and edit them :) If you want to contribute, we normally discuss this through Discord!