Dossier
Fermisense
Coverage of Fermisense in the Nexus archive.
- A $500 RL fine-tune of a 9B open model beat frontier models on catalog review
A $500 reinforcement learning fine-tune of a 9B open model outperformed frontier models in a catalog review. The results were discussed on Hacker News, where the article received 24 points and 5 comments.