Text Generation
Transformers
Safetensors
cohere2
conversational

Finetuning instruction & GPU advise

#2
by millionwords - opened

Can you give me instructions & code to finetune the model?
Secondly, how many GPUs would i need to finetune it?

Cohere Labs org

@millionwords You can find some instructions on fine-tuning in the HF docs -- https://huggingface.co/blog/dvgodoy/fine-tuning-llm-hugging-face

For the amount of GPUs needed, it depends on many things like the VRAM available in the GPUs you are using, the batch size, if you are doing full fine-tuning or LoRA/QLoRA, etc.

alexrs changed discussion status to closed

Sign up or log in to comment