finetuning
#2
by
HassanStar
- opened
Any explanation how to finetune it?
Thanks
@HassanStar Did you manage to figure out how to fine-tune?
Hey, for finetuning we recommend using https://github.com/huggingface/trl or https://github.com/huggingface/alignment-handbook :)
Can you provide a pure-python PEFT adapter.py script for each of the SmolLM2 models so that the model can be downloaded and finetuned on local PC using the PEFT/Adapter method instead of suggesting that people must finetune the whole model weight sets?
You can find a fine-tuning script with PEFT here: https://github.com/huggingface/smollm/tree/main/finetune you can adapt it for an SFT dataset