Contrastive Activation Addition (CAA) is a method for steering language models by modifying activations during forward passes. It outperforms traditional methods and provides insights into how concepts are represented in Large Language Models (LLMs).

https://arxiv.org/abs//2312.06681

YouTube: https://www.youtube.com/@ArxivPapers

TikTok: https://www.tiktok.com/@arxiv_papers

Apple Podcasts: https://podcasts.apple.com/us/podcast/arxiv-papers/id1692476016

Spotify: https://podcasters.spotify.com/pod/show/arxiv-papers