Get our free extension to see links to code for papers anywhere online!

Chrome logo Add to Chrome

Firefox logo Add to Firefox
Picture for Sergey Levine

Sergey Levine

Benchmarks for Deep Off-Policy Evaluation


Mar 30, 2021
Justin Fu, Mohammad Norouzi, Ofir Nachum, George Tucker, Ziyu Wang, Alexander Novikov, Mengjiao Yang, Michael R. Zhang, Yutian Chen, Aviral Kumar, Cosmin Paduraru, Sergey Levine, Tom Le Paine

Figure 1 for Benchmarks for Deep Off-Policy Evaluation
Figure 2 for Benchmarks for Deep Off-Policy Evaluation
Figure 3 for Benchmarks for Deep Off-Policy Evaluation
Figure 4 for Benchmarks for Deep Off-Policy Evaluation
* ICLR 2021 paper. Policies and evaluation code are available at https://github.com/google-research/deep_ope 
  Access Paper or Ask Questions

Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots


Mar 26, 2021
Zhongyu Li, Xuxin Cheng, Xue Bin Peng, Pieter Abbeel, Sergey Levine, Glen Berseth, Koushil Sreenath

Figure 1 for Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots
Figure 2 for Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots
Figure 3 for Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots
Figure 4 for Reinforcement Learning for Robust Parameterized Locomotion Control of Bipedal Robots
* To appear on 2021 International Conference on Robotics and Automation (ICRA 2021) 
  Access Paper or Ask Questions

Policy Information Capacity: Information-Theoretic Measure for Task Complexity in Deep Reinforcement Learning


Mar 23, 2021
Hiroki Furuta, Tatsuya Matsushima, Tadashi Kozuno, Yutaka Matsuo, Sergey Levine, Ofir Nachum, Shixiang Shane Gu

Figure 1 for Policy Information Capacity: Information-Theoretic Measure for Task Complexity in Deep Reinforcement Learning
Figure 2 for Policy Information Capacity: Information-Theoretic Measure for Task Complexity in Deep Reinforcement Learning
Figure 3 for Policy Information Capacity: Information-Theoretic Measure for Task Complexity in Deep Reinforcement Learning
Figure 4 for Policy Information Capacity: Information-Theoretic Measure for Task Complexity in Deep Reinforcement Learning

  Access Paper or Ask Questions

Replacing Rewards with Examples: Example-Based Policy Search via Recursive Classification


Mar 23, 2021
Benjamin Eysenbach, Sergey Levine, Ruslan Salakhutdinov

Figure 1 for Replacing Rewards with Examples: Example-Based Policy Search via Recursive Classification
Figure 2 for Replacing Rewards with Examples: Example-Based Policy Search via Recursive Classification
Figure 3 for Replacing Rewards with Examples: Example-Based Policy Search via Recursive Classification
Figure 4 for Replacing Rewards with Examples: Example-Based Policy Search via Recursive Classification
* Website with videos and code: https://ben-eysenbach.github.io/rce 
  Access Paper or Ask Questions

Maximum Entropy RL (Provably) Solves Some Robust RL Problems


Mar 10, 2021
Benjamin Eysenbach, Sergey Levine

Figure 1 for Maximum Entropy RL (Provably) Solves Some Robust RL Problems
Figure 2 for Maximum Entropy RL (Provably) Solves Some Robust RL Problems
Figure 3 for Maximum Entropy RL (Provably) Solves Some Robust RL Problems
Figure 4 for Maximum Entropy RL (Provably) Solves Some Robust RL Problems
* Blog post and videos: https://bair.berkeley.edu/blog/2021/03/10/maxent-robust-rl/. arXiv admin note: text overlap with arXiv:1910.01913 
  Access Paper or Ask Questions

PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning


Feb 24, 2021
Angelos Filos, Clare Lyle, Yarin Gal, Sergey Levine, Natasha Jaques, Gregory Farquhar

Figure 1 for PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning
Figure 2 for PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning
Figure 3 for PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning
Figure 4 for PsiPhi-Learning: Reinforcement Learning with Demonstrations using Successor Features and Inverse Temporal Difference Learning

  Access Paper or Ask Questions

COMBO: Conservative Offline Model-Based Policy Optimization


Feb 16, 2021
Tianhe Yu, Aviral Kumar, Rafael Rafailov, Aravind Rajeswaran, Sergey Levine, Chelsea Finn

Figure 1 for COMBO: Conservative Offline Model-Based Policy Optimization
Figure 2 for COMBO: Conservative Offline Model-Based Policy Optimization
Figure 3 for COMBO: Conservative Offline Model-Based Policy Optimization
Figure 4 for COMBO: Conservative Offline Model-Based Policy Optimization

  Access Paper or Ask Questions

Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation


Feb 16, 2021
Justin Fu, Sergey Levine

Figure 1 for Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation
Figure 2 for Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation
Figure 3 for Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation
Figure 4 for Offline Model-Based Optimization via Normalized Maximum Likelihood Estimation

  Access Paper or Ask Questions

How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned


Feb 04, 2021
Julian Ibarz, Jie Tan, Chelsea Finn, Mrinal Kalakrishnan, Peter Pastor, Sergey Levine

Figure 1 for How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned
Figure 2 for How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned
Figure 3 for How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned
Figure 4 for How to Train Your Robot with Deep Reinforcement Learning; Lessons We've Learned
* Journal of Robotics Research (IJRR), February 2021 
  Access Paper or Ask Questions

SimGAN: Hybrid Simulator Identification for Domain Adaptation via Adversarial Reinforcement Learning


Jan 15, 2021
Yifeng Jiang, Tingnan Zhang, Daniel Ho, Yunfei Bai, C. Karen Liu, Sergey Levine, Jie Tan

Figure 1 for SimGAN: Hybrid Simulator Identification for Domain Adaptation via Adversarial Reinforcement Learning
Figure 2 for SimGAN: Hybrid Simulator Identification for Domain Adaptation via Adversarial Reinforcement Learning
Figure 3 for SimGAN: Hybrid Simulator Identification for Domain Adaptation via Adversarial Reinforcement Learning
Figure 4 for SimGAN: Hybrid Simulator Identification for Domain Adaptation via Adversarial Reinforcement Learning
* Submitted to ICRA 2021 
  Access Paper or Ask Questions

Evolving Reinforcement Learning Algorithms


Jan 08, 2021
John D. Co-Reyes, Yingjie Miao, Daiyi Peng, Esteban Real, Sergey Levine, Quoc V. Le, Honglak Lee, Aleksandra Faust

Figure 1 for Evolving Reinforcement Learning Algorithms
Figure 2 for Evolving Reinforcement Learning Algorithms
Figure 3 for Evolving Reinforcement Learning Algorithms
Figure 4 for Evolving Reinforcement Learning Algorithms

  Access Paper or Ask Questions

Model-Based Visual Planning with Self-Supervised Functional Distances


Dec 30, 2020
Stephen Tian, Suraj Nair, Frederik Ebert, Sudeep Dasari, Benjamin Eysenbach, Chelsea Finn, Sergey Levine

Figure 1 for Model-Based Visual Planning with Self-Supervised Functional Distances
Figure 2 for Model-Based Visual Planning with Self-Supervised Functional Distances
Figure 3 for Model-Based Visual Planning with Self-Supervised Functional Distances
Figure 4 for Model-Based Visual Planning with Self-Supervised Functional Distances

  Access Paper or Ask Questions

ViNG: Learning Open-World Navigation with Visual Goals


Dec 17, 2020
Dhruv Shah, Benjamin Eysenbach, Gregory Kahn, Nicholas Rhinehart, Sergey Levine

Figure 1 for ViNG: Learning Open-World Navigation with Visual Goals
Figure 2 for ViNG: Learning Open-World Navigation with Visual Goals
Figure 3 for ViNG: Learning Open-World Navigation with Visual Goals
Figure 4 for ViNG: Learning Open-World Navigation with Visual Goals

  Access Paper or Ask Questions

Variable-Shot Adaptation for Online Meta-Learning


Dec 14, 2020
Tianhe Yu, Xinyang Geng, Chelsea Finn, Sergey Levine

Figure 1 for Variable-Shot Adaptation for Online Meta-Learning
Figure 2 for Variable-Shot Adaptation for Online Meta-Learning
Figure 3 for Variable-Shot Adaptation for Online Meta-Learning
Figure 4 for Variable-Shot Adaptation for Online Meta-Learning
* First two authors contribute equally 
  Access Paper or Ask Questions

WILDS: A Benchmark of in-the-Wild Distribution Shifts


Dec 14, 2020
Pang Wei Koh, Shiori Sagawa, Henrik Marklund, Sang Michael Xie, Marvin Zhang, Akshay Balsubramani, Weihua Hu, Michihiro Yasunaga, Richard Lanas Phillips, Sara Beery, Jure Leskovec, Anshul Kundaje, Emma Pierson, Sergey Levine, Chelsea Finn, Percy Liang

Figure 1 for WILDS: A Benchmark of in-the-Wild Distribution Shifts
Figure 2 for WILDS: A Benchmark of in-the-Wild Distribution Shifts
Figure 3 for WILDS: A Benchmark of in-the-Wild Distribution Shifts
Figure 4 for WILDS: A Benchmark of in-the-Wild Distribution Shifts

  Access Paper or Ask Questions

Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning


Dec 08, 2020
Mohammad Babaeizadeh, Mohammad Taghi Saffar, Danijar Hafner, Harini Kannan, Chelsea Finn, Sergey Levine, Dumitru Erhan

Figure 1 for Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning
Figure 2 for Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning
Figure 3 for Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning
Figure 4 for Models, Pixels, and Rewards: Evaluating Design Trade-offs in Visual Model-Based Reinforcement Learning

  Access Paper or Ask Questions

Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design


Dec 03, 2020
Michael Dennis, Natasha Jaques, Eugene Vinitsky, Alexandre Bayen, Stuart Russell, Andrew Critch, Sergey Levine

Figure 1 for Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design
Figure 2 for Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design
Figure 3 for Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design
Figure 4 for Emergent Complexity and Zero-shot Transfer via Unsupervised Environment Design

  Access Paper or Ask Questions

Parrot: Data-Driven Behavioral Priors for Reinforcement Learning


Nov 19, 2020
Avi Singh, Huihan Liu, Gaoyue Zhou, Albert Yu, Nicholas Rhinehart, Sergey Levine

Figure 1 for Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
Figure 2 for Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
Figure 3 for Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
Figure 4 for Parrot: Data-Driven Behavioral Priors for Reinforcement Learning
* First two authors contributed equally. Project website: https://sites.google.com/view/parrot-rl 
  Access Paper or Ask Questions

C-Learning: Learning to Achieve Goals via Recursive Classification


Nov 17, 2020
Benjamin Eysenbach, Ruslan Salakhutdinov, Sergey Levine

Figure 1 for C-Learning: Learning to Achieve Goals via Recursive Classification
Figure 2 for C-Learning: Learning to Achieve Goals via Recursive Classification
Figure 3 for C-Learning: Learning to Achieve Goals via Recursive Classification
Figure 4 for C-Learning: Learning to Achieve Goals via Recursive Classification
* Project website: https://ben-eysenbach.github.io/c_learning/ 
  Access Paper or Ask Questions

Reinforcement Learning with Videos: Combining Offline Observations with Interaction


Nov 12, 2020
Karl Schmeckpeper, Oleh Rybkin, Kostas Daniilidis, Sergey Levine, Chelsea Finn

Figure 1 for Reinforcement Learning with Videos: Combining Offline Observations with Interaction
Figure 2 for Reinforcement Learning with Videos: Combining Offline Observations with Interaction
Figure 3 for Reinforcement Learning with Videos: Combining Offline Observations with Interaction
Figure 4 for Reinforcement Learning with Videos: Combining Offline Observations with Interaction

  Access Paper or Ask Questions

Continual Learning of Control Primitives: Skill Discovery via Reset-Games


Nov 10, 2020
Kelvin Xu, Siddharth Verma, Chelsea Finn, Sergey Levine

Figure 1 for Continual Learning of Control Primitives: Skill Discovery via Reset-Games
Figure 2 for Continual Learning of Control Primitives: Skill Discovery via Reset-Games
Figure 3 for Continual Learning of Control Primitives: Skill Discovery via Reset-Games
Figure 4 for Continual Learning of Control Primitives: Skill Discovery via Reset-Games
* To appear at NeurIPS 2020 
  Access Paper or Ask Questions

Amortized Conditional Normalized Maximum Likelihood


Nov 05, 2020
Aurick Zhou, Sergey Levine

Figure 1 for Amortized Conditional Normalized Maximum Likelihood
Figure 2 for Amortized Conditional Normalized Maximum Likelihood
Figure 3 for Amortized Conditional Normalized Maximum Likelihood
Figure 4 for Amortized Conditional Normalized Maximum Likelihood

  Access Paper or Ask Questions

Rearrangement: A Challenge for Embodied AI


Nov 03, 2020
Dhruv Batra, Angel X. Chang, Sonia Chernova, Andrew J. Davison, Jia Deng, Vladlen Koltun, Sergey Levine, Jitendra Malik, Igor Mordatch, Roozbeh Mottaghi, Manolis Savva, Hao Su

Figure 1 for Rearrangement: A Challenge for Embodied AI
Figure 2 for Rearrangement: A Challenge for Embodied AI
Figure 3 for Rearrangement: A Challenge for Embodied AI
Figure 4 for Rearrangement: A Challenge for Embodied AI
* Authors are listed in alphabetical order 
  Access Paper or Ask Questions

COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning


Oct 27, 2020
Avi Singh, Albert Yu, Jonathan Yang, Jesse Zhang, Aviral Kumar, Sergey Levine

Figure 1 for COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
Figure 2 for COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
Figure 3 for COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
Figure 4 for COG: Connecting New Skills to Past Experience with Offline Reinforcement Learning
* Accepted to CoRL 2020. Source code and videos available at https://sites.google.com/view/cog-rl 
  Access Paper or Ask Questions

Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning


Oct 27, 2020
Aviral Kumar, Rishabh Agarwal, Dibya Ghosh, Sergey Levine

Figure 1 for Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning
Figure 2 for Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning
Figure 3 for Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning
Figure 4 for Implicit Under-Parameterization Inhibits Data-Efficient Deep Reinforcement Learning
* Pre-print. First two authors contributed equally 
  Access Paper or Ask Questions

Conservative Safety Critics for Exploration


Oct 27, 2020
Homanga Bharadhwaj, Aviral Kumar, Nicholas Rhinehart, Sergey Levine, Florian Shkurti, Animesh Garg

Figure 1 for Conservative Safety Critics for Exploration
Figure 2 for Conservative Safety Critics for Exploration
Figure 3 for Conservative Safety Critics for Exploration
Figure 4 for Conservative Safety Critics for Exploration
* Preprint. Under review 
  Access Paper or Ask Questions

$γ$-Models: Generative Temporal Difference Learning for Infinite-Horizon Prediction


Oct 27, 2020
Michael Janner, Igor Mordatch, Sergey Levine

Figure 1 for $γ$-Models: Generative Temporal Difference Learning for Infinite-Horizon Prediction
Figure 2 for $γ$-Models: Generative Temporal Difference Learning for Infinite-Horizon Prediction
Figure 3 for $γ$-Models: Generative Temporal Difference Learning for Infinite-Horizon Prediction
Figure 4 for $γ$-Models: Generative Temporal Difference Learning for Infinite-Horizon Prediction
* NeurIPS 2020. Project page at: https://people.eecs.berkeley.edu/~janner/gamma-models/ 
  Access Paper or Ask Questions

One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL


Oct 27, 2020
Saurabh Kumar, Aviral Kumar, Sergey Levine, Chelsea Finn

Figure 1 for One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL
Figure 2 for One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL
Figure 3 for One Solution is Not All You Need: Few-Shot Extrapolation via Structured MaxEnt RL
* Accepted at NeurIPS 2020 
  Access Paper or Ask Questions

OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning


Oct 27, 2020
Anurag Ajay, Aviral Kumar, Pulkit Agrawal, Sergey Levine, Ofir Nachum

Figure 1 for OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
Figure 2 for OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
Figure 3 for OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
Figure 4 for OPAL: Offline Primitive Discovery for Accelerating Offline Reinforcement Learning
* https://sites.google.com/view/opal-iclr 
  Access Paper or Ask Questions

MELD: Meta-Reinforcement Learning from Images via Latent State Models


Oct 26, 2020
Tony Z. Zhao, Anusha Nagabandi, Kate Rakelly, Chelsea Finn, Sergey Levine

Figure 1 for MELD: Meta-Reinforcement Learning from Images via Latent State Models
Figure 2 for MELD: Meta-Reinforcement Learning from Images via Latent State Models
Figure 3 for MELD: Meta-Reinforcement Learning from Images via Latent State Models
Figure 4 for MELD: Meta-Reinforcement Learning from Images via Latent State Models
* Accepted to CoRL 2020. Supplementary material at https://sites.google.com/view/meld-lsm/home . 16 pages, 19 figures 
  Access Paper or Ask Questions