Sergey Levine
Associate Professor
Develops deep reinforcement learning algorithms for decision making and control in robots.
Sergey Levine's lab at UC Berkeley works on reinforcement learning. 164 papers since 2023, 151 at top venues, mostly ICLR and ICML. 45 alumni, most often to Google and Google DeepMind. Applications close .
At a glance
- papers, 2023–2026
- 43/ year
- at T1 venues
- 92%
- PhD students
- 0
- MS students
- 0
- alumni listed
- 45
Papers by yearfrom DBLP, 2023–2026
- 18 papers2026 · T1 15 · collecting
- 43 papers2025 · T1 39
- 54 papers2024 · T1 51
- 49 papers2023 · T1 46
T1 venuesOtherVenues: ICLR, ICML, NeurIPS, CoRL, ICRA
Research focuspaper titles that use each topic's words
- Reinforcement learning90 papers
- deep reinforcement2 papers
- decision making1 paper
- develops deep0 papers
- learning algorithms0 papers
Paper list
From DBLP.
202618 papers
- IEEE Robotics Autom. Lett.CAST: Counterfactual Labels Improve Instruction Following in Vision-Language-Action Models (opens in a new tab)
- AAAI (T1)Cliqueformer: Model-Based Optimization with Structured Transformers (opens in a new tab)
- ICLR (T1)Decoupled Q-Chunking (opens in a new tab)
- ICLR (T1)Dual Goal Representations (opens in a new tab)
- ICLR (T1)Intention-Conditioned Flow Occupancy Models (opens in a new tab)
- ICRA (T1)Learning Affordances at Inference-Time for Vision-Language-Action Models (opens in a new tab)
- IEEE Robotics Autom. Lett.Learning to Drive Anywhere With Model-Based Reannotation (opens in a new tab)
- ICLR (T1)Learning to Reason without External Rewards (opens in a new tab)
- ICLR (T1)Leveraging Discrete Function Decomposability for Scientific Design (opens in a new tab)
- ICRA (T1)OmniVLA: An Omni-Modal Vision-Language-Action Model for Robot Navigation (opens in a new tab)
- ICLR (T1)Q-Learning with Adjoint Matching (opens in a new tab)
- ICLR (T1)RLAC: Reinforcement Learning with Adversarial Critic for Free-Form Generation Tasks (opens in a new tab)
202543 papers
- NeurIPS (T1)A Stable Whitening Optimizer for Efficient Neural Network Training (opens in a new tab)
- ICLR (T1)Adding Conditional Control to Diffusion Models with Reinforcement Learning (opens in a new tab)
- ICML (T1)Behavioral Exploration: Learning to Explore via In-Context Adaptation (opens in a new tab)
- ICRA (T1)Beyond Sight: Finetuning Generalist Robot Policies with Heterogeneous Sensors via Language Grounding (opens in a new tab)
- ICRA (T1)Commonsense Reasoning for Legged Robot Adaptation with Vision-Language Models (opens in a new tab)
- NeurIPS (T1)Compute-Optimal Scaling for Value-Based Deep RL (opens in a new tab)
- NeurIPS (T1)Consistently Simulating Human Personas with Multi-Turn Reinforcement Learning (opens in a new tab)
- NeurIPS (T1)Derivative-Free Guidance in Continuous and Discrete Diffusion Models with Soft Value-based Decoding (opens in a new tab)
- ICLR (T1)Digi-Q: Learning VLM Q-Value Functions for Training Device-Control Agents (opens in a new tab)
- ICLR (T1)Efficient Online Reinforcement Learning Fine-Tuning Need Not Retain Offline Data (opens in a new tab)
- ICLR (T1)Fine-Tuning Discrete Diffusion Models via Reward Optimization with Applications to DNA and Protein Design (opens in a new tab)
- ICML (T1)Flow Q-Learning (opens in a new tab)
202454 papers
- ICML (T1)ArCHer: Training Language Model Agents via Hierarchical Multi-Turn RL (opens in a new tab)
- CoRL (T1)Autonomous Improvement of Instruction Following Skills via Foundation Models (opens in a new tab)
- NeurIPS (T1)Bridging Model-Based Optimization and Generative Modeling via Conservative Fine-Tuning of Diffusion Models (opens in a new tab)
- ICML (T1)Chain of Code: Reasoning with a Language Model-Augmented Code Emulator (opens in a new tab)
- RLJD5RL: Diverse Datasets for Data-Driven Deep Reinforcement Learning (opens in a new tab)
- ICLR (T1)Deep Neural Networks Tend To Extrapolate Predictably (opens in a new tab)
- AAMAS (T1)Defining Deception in Decision Making (opens in a new tab)
- NeurIPS (T1)Designing Cell-Type-Specific Promoter Sequences Using Conservative Model-Based Optimization (opens in a new tab)
- NeurIPS (T1)DigiRL: Training In-The-Wild Device-Control Agents with Autonomous Reinforcement Learning (opens in a new tab)
- RSS (T1)DROID: A Large-Scale In-The-Wild Robot Manipulation Dataset (opens in a new tab)
- CoRL (T1)Evaluating Real-World Robot Manipulation Policies in Simulation (opens in a new tab)
- ICML (T1)Feedback Efficient Online Fine-Tuning of Diffusion Models (opens in a new tab)
202349 papers
- ICML (T1)A Connection between One-Step RL and Critic Regularization in Reinforcement Learning (opens in a new tab)
- NeurIPS (T1)Accelerating Exploration with Unlabeled Prior Data (opens in a new tab)
- CoRL (T1)Action-Quantized Offline Reinforcement Learning for Robotic Skill Learning (opens in a new tab)
- ICML (T1)Adversarial Policies Beat Superhuman Go AIs (opens in a new tab)
- ICLR (T1)Bitrate-Constrained DRO: Beyond Worst Case Robustness To Unknown Group Shifts (opens in a new tab)
- IROS (T1)Bootstrapping Adaptive Human-Machine Interfaces with Offline Reinforcement Learning (opens in a new tab)
- CoRL (T1)BridgeData V2: A Dataset for Robot Learning at Scale (opens in a new tab)
- NeurIPS (T1)Cal-QL: Calibrated Offline RL Pre-Training for Efficient Online Fine-Tuning (opens in a new tab)
- ICLR (T1)Confidence-Conditioned Value Functions for Offline Reinforcement Learning (opens in a new tab)
- L4DCContrastive Example-Based Control (opens in a new tab)
- RSS (T1)Deep RL at Scale: Sorting Waste in Office Buildings with a Fleet of Mobile Manipulators (opens in a new tab)
- RSS (T1)Demonstrating A Walk in the Park: Learning to Walk in 20 Minutes With Model-Free Reinforcement Learning (opens in a new tab)
Where alumni went45 listed, 75% to industry
Industry 27Academia 9
Most often to
- Google DeepMind
- OpenAI
Also to
- University of Washington
- Cornell University
- Princeton University
Apply to UC BerkeleyFall 2027
UC Berkeley EECS Computer Science PhD
- Round
- Single round
- Opens
- —
- GRE
- not required
- English
- TOEFL/IELTS (non-English-country degrees)
- Fee
- See program page
Confirmed on the program page.
Similar labsnearest by research summary
- GBEdinburgh
Viacheslav Borovitskiy
- Probabilistic & causal inference
- CNSJTU
Bo Zhao
- Multimodal & grounded language
- SGNTU
Stefano V. Albrecht
- Reinforcement learning
- USUMD
Kaiqing Zhang
- Reasoning & neuro-symbolic AI
- CNZhejiang
Ruijie Quan
- Vision-language & multimodal
- CNTsinghua
Shiji Song
- Learning theory & optimization
Your notes
Save this lab to keep notes.
Data
Checked against the lab homepage.
- Collected
- Verified
- Papers
- DBLP
- Members
- Lab homepage
- Alumni
- Lab homepage
Tiers are venue classifications, not lab rankings.
UC Berkeleycloses