Comment

Comment by rybosome

I find this compelling “merely” as an engineering challenge, but I’m curious if you’ve observed an actual practical benefit?

By that I mean, have you found a case where a fine-tuned, small model is outperformed by stuffing a heavily tuned larger model in its place?

parent rybosome 9 days ago

Discussion

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton