Ask HN: Ethics on fine-tuning LLMs on books and papers

Posted by zz_zac on Thursday May 18, 2023 @06:19PM
from the ask-hn dept.

I want to get started on fine-tuning LLMs to emulate certain styles of writing (e.g. online posts, blogs, etc.) - but I don't quite understand the ethical and/or legal implications.

What are some ethical ways of sourcing such data? Thanks!

( 2 comments | on HN )

Re:Ask HN: Ethics on fine-tuning LLMs on books and papers by PaulHoule on Thursday May 18, 2023 @06:33PM (#35995139)

I've talked to people that think there could be a problem with my smart RSS reader that retains the content of RSS feeds and trains a classification model to predict if I like the article and not even generate anything.

[ Parent | on HN ]

Re:Ask HN: Ethics on fine-tuning LLMs on books and papers by ftxbro on Thursday May 18, 2023 @06:20PM (#35994984)

I think ethical or legal implications might depend what you do with the LLMs.

[ Parent | on HN ]