2310.01405.pdf
arxiv.org · 7,815 words · saved by 2 readers
N/A
R EPRESENTATION E NGINEERING : A T OP -D OWN A PPROACH TO AI T RANSPARENCY Andy Zou1,2 , Long Phan∗1 , Sarah Chen∗1,4 , James Campbell∗7 , Phillip Guo∗6 , Richard Ren∗8 , Alexander Pan3 , Xuwang Yin1 , Mantas Mazeika1,9 , Ann-Kathrin Dombrowski1 , Shashwat Goel1 , Nathaniel Li1,3 , Michael J. Byun4 , Zifan Wang1 , Alex Mallen5 , Steven Basart1 ,…
saved by
related reading
- 2310.01405arxiv.org
- 2307.12950.pdfarxiv.org
- [2310.01405] Representation Engineering: A Top-Down Approach to AI Transparencyarxiv.org
- [2210.13382] Emergent World Representations: Exploring a Sequence Model Trained on a Synthetic Taskarxiv.org
- 2409.02908arxiv.org
- A Survey on Sparse Autoencoders: Interpreting the Internal Mechanisms of Large Language Modelsarxiv.org
- 2403.09611.pdfarxiv.org
- 2312.06942arxiv.org
- 2308.03958arxiv.org
- Fact Finding: Attempting to Reverse-Engineer Factual Recall on the Neuron Level (Post 1) — AI Alignment Forumalignmentforum.org
- 2305.18654arxiv.org
- 2309.07311.pdfarxiv.org