Picture for Víctor Abia Alonso

Víctor Abia Alonso

Reinforcement Learning Fine-tuning of Language Models is Biased Towards More Extractable Features

Add code
Nov 07, 2023
Viaarxiv icon