AxBench: Steering LLMs? Even Simple Baselines Outperform Sparse Autoencoders

Add code
Jan 29, 2025

Share this with someone who'll enjoy it:

View paper onarxiv icon

Share this with someone who'll enjoy it: