Slacker News Slacker News logo featuring a lazy sloth with a folded newspaper hat
  • top
  • new
  • show
  • ask
  • jobs
Library
← Back to context

Comment by gruez

5 months ago

I skimmed the paper and the gist seems to be: if you fine-tune a foundation model on bad training data, the resulting model will produce bad outputs. That seems... expected? This makes as much sense as "if you add vulnerable libraries to your app, your app will be vulnerable". I'm not sure how this can turn into an actual attack though.

0 comments

gruez

Reply

No comments yet

Contribute on Hacker News ↗

Slacker News

Product

  • API Reference
  • Hacker News RSS
  • Source on GitHub

Community

  • Support Ukraine
  • Equal Justice Initiative
  • GiveWell Charities