Ben Millwood🔸 comments on Ben Millwood’s Quick takes

[Ben Millwood🔸]

I made this into two posts, my first LessWrong posts:

• Keeping content out of LLM training datasets

• Should we exclude alignment research from LLM training datasets?