Paper
Verbalizable Representations Form a Global Workspace in Language Models
This paper introduces the Jacobian lens (J-lens), a new interpretability technique that identifies which internal representations in a language model are ver...
Related papers
36
-
Anthropic
-
MATS
-
Theorem
-
UK AISI
-
Alignment Research Center
-
Anthropic
-
Truthful AI
-
University of California, Berkeley
-
Warsaw University of Technology
-
Center for AI Safety
-
Center for AI Standards and Innovation
-
Co-affiliated with Carnegie Mellon University
-
UK Artificial Intelligence Security Institute
-
AI Futures Project
-
Apollo Research
-
Center for AI Safety
-
Mila
-
Model Evaluation & Threat Research
-
Redwood Research
-
Truthful AI
-
UK AI Security Institute
-
University of California, Berkeley
-
University of Montreal
Looking for more?
Open the arXiv / PDF link for the full paper. Abstract and authors appear here when enrichment has run.