Friday, March 15, 2024

How to properly finetune an LLM when you only have text, not formatted Q&A

title says it all, but basically how would did openai, llama, and mistral train on so much text that can't be manually formatted?

how could we finetune them without having to manually and painstakingly format our text into json?


Comments URL: https://news.ycombinator.com/item?id=39719148

Points: 1

# Comments: 0



from Hacker News: Newest https://ift.tt/PNB1VFt

0 comments:

Post a Comment