How to think about creating a dataset for LLM fine-tuning evaluation

Re:How to think about creating a dataset for LLM fine-tuning... by msp26 on Thursday June 27, 2024 @07:49AM (#40809467)

For tasks like data extraction, are people doing full finetunes or training a LoRA?

[ Parent | on HN ]

Re:How to think about creating a dataset for LLM fine-tuning... by clarionbell on Thursday June 27, 2024 @08:01AM (#40809556)

Unless you have GPUs available, LoRa is the only accessible option. That being said, if your task is simple enough, you can skip the problem entirely and just pick a small base model.

[ Parent | on HN ]

Re:How to think about creating a dataset for LLM fine-tuning... by swalsh on Thursday June 27, 2024 @08:03AM (#40809570)

LoRA is perfect for that. Where it fails is when you want to teach a new domain.

[ Parent | on HN ]