Picture for Luke Marks

Luke Marks

Interpreting Reward Models in RLHF-Tuned Language Models Using Sparse Autoencoders

Add code
Oct 12, 2023
Viaarxiv icon