Hacker News new | ask | show | jobs
by _lvbh 903 days ago
I did something similar for my extended essay (IB class of 2023) last year. (Evaluated the effectiveness of fine tuning at removing censorship from training phase)

Didn’t do too well. Make sure to focus on the theory & cite lots of papers. Don’t focus on practicals (at least that was the advice I got from my school)

1 comments

Oh, interesting. Obviously, this is a decision I have to make for myself -- but would you recommend I choose a different path? It's not too late yet.

I have been concerned that because ML is changing rapidly, it may be challenging to find cite-able sources.

To be honest, I have no idea. All I know is that you definitely need quite a few cite-able sources, if for any reason, just to tick boxes