6.3k citations
- University of California, BerkeleyUS18 papers
- Google (United States)US11 papers
- Stanford UniversityUS8 papers
- Carnegie Mellon UniversityUS6 papers
- Berkeley CollegeUS4 papers
- Uber AI (United States)US4 papers
- ETH ZurichCH3 papers
- Institut national de recherche en sciences et technologies du numériqueFR3 papers
- McGill UniversityCA3 papers
- Microsoft (United States)US3 papers
- Mila - Quebec Artificial Intelligence InstituteCA3 papers
- The University of Texas at AustinUS3 papers
18 papers · 1 filter
Extracting Training Data from Large Language Models
Nicholas Carlini, Florian Tramer, Eric Wallace +9
It has become common to publish large (billion parameter) language models that have been trained on private datasets. This paper demonstrates that in such settings, an adversary ca…
Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images
Rewon Child
We present a hierarchical VAE that, for the first time, generates samples quickly while outperforming the PixelCNN in log-likelihood on all natural image benchmarks. We begin by ob…
Behavior Priors for Efficient Reinforcement Learning
Dhruva Tirumala, Alexandre Galashov, Hyeonwoo Noh +8
As we deploy reinforcement learning agents to solve increasingly challenging problems, methods that allow us to inject prior knowledge about the structure of the world and effectiv…
Emergent Social Learning via Multi-agent Reinforcement Learning
Kamal Ndousse, Douglas Eck, Sergey Levine +1
Social learning is a key component of human and animal intelligence. By taking cues from the behavior of experts in their environment, social learners can acquire sophisticated beh…
Measurement in AI Policy: Opportunities and Challenges
Saurabh Mishra, Jack Clark, C. Raymond Perrault
As artificial intelligence increasingly influences our world, it becomes crucial to assess its technical progress and societal impact. This paper surveys problems and opportunities…
Phasic Policy Gradient
Karl Cobbe, Jacob Hilton, Oleg Klimov +1
We introduce Phasic Policy Gradient (PPG), a reinforcement learning framework which modifies traditional on-policy actor-critic methods by separating policy and value function trai…