Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct von Ackerman, Christopher, Panickssery, Nina

Volltext bestellen

Understanding Jailbreak Success: A Study of Latent Space Dynamics in Large Language Models von Ball, Sarah, Kreuter, Frauke, Panickssery, Nina

Volltext bestellen

Inspection and Control of Self-Generated-Text Recognition Ability in Llama3-8b-Instruct von Ackerman, Christopher, Panickssery, Nina

Volltext

Refusal in Language Models Is Mediated by a Single Direction von Arditi, Andy, Obeso, Oscar, Syed, Aaquib, Paleka, Daniel, Panickssery, Nina, Gurnee, Wes, Nanda, Neel

Volltext bestellen

Steering Llama 2 via Contrastive Activation Addition von Panickssery, Nina, Gabrieli, Nick, Schulz, Julian, Tong, Meg, Hubinger, Evan, Turner, Alexander Matt

Volltext bestellen

Refusal in Language Models Is Mediated by a Single Direction von Arditi, Andy, Obeso, Oscar, Aaquib Syed, Paleka, Daniel, Panickssery, Nina, Gurnee, Wes, Nanda, Neel

Volltext

Steering Llama 2 via Contrastive Activation Addition von Panickssery, Nina, Gabrieli, Nick, Schulz, Julian, Tong, Meg, Hubinger, Evan, Turner, Alexander Matt

Volltext