Dibya Ghosh

I'm a final-year PhD student at UC Berkeley training large models using reinforcement learning. Previously, I was at Google Brain, Montréal. Even earlier, I did my undergrad at UC Berkeley.

See my Google Scholar for published work and blog for more fun.

My first name is pronounced "Dibbo". Reach me at dibya (at) berkeley (dot) edu.

Find my old website here

Selected Publications

Annotation Bootstrapping: A Self-Reinforcing Approach to Visual Pre-Training . Preprint
Dibya Ghosh, Sergey Levine.
Octo: An Open-Source Generalist Robot Policy. RSS 2024
Dibya Ghosh, Homer Walke, Karl Pertsch, Kevin Black, Oier Mees, et. al.
Reinforcement Learning from Passive Data via Latent Intentions. ICML 2023 (Oral).
Dibya Ghosh, Chethan Bhateja, Sergey Levine.
Offline RL Policies Should be Trained to be Adaptive . ICML 2022 (Oral).
Dibya Ghosh, Anurag Ajay, Pulkit Agrawal, Sergey Levine.
Why Generalization in RL is Difficult: Epistemic POMDPs and Implicit Partial Observability. NeurIPS 2021.
Dibya Ghosh, Jad Rahme, Aviral Kumar, Amy Zhang, Ryan P. Adams, Sergey Levine.
Learning to Reach Goals via Iterated Supervised Learning . ICLR 2021 (Oral).
Dibya Ghosh, Abhishek Gupta, Ashwin Reddy, Justin Fu, Coline Devin, Ben Eysenbach, Sergey Levine.
See all ▼

Robotic Offline RL from Internet Videos via Value-Function Pre-Training. Preprint.
Chethan Bhateja*, Derek Guo*, Dibya Ghosh*, Anikait Singh, Manan Tomar, Quan Vuong, Yevgen Chebotar, Sergey Levine, Aviral Kumar.
Hiql: Offline goal-conditioned rl with latent states as actions. NeurIPS 2023.
Seohong Park, Dibya Ghosh, Ben Eysenbach, Sergey Levine.
Accelerating Exploration with Unlabeled Prior Data . NeurIPS 2023.
Qiyang Li, Jason Zhang, Dibya Ghosh, Amy Zhang, Sergey Levine.
Distributionally Adaptive Meta Reinforcement Learning . NeurIPS 2022.
Anurag Ajay, Abhishek Gupta, Dibya Ghosh, Sergey Levine, Pulkit Agrawal.
Implicit Under-Parameterization Inhibits Data-Efficient Deep RL . ICLR 2021.
Aviral Kumar, Rishabh Agarwal, Dibya Ghosh, Sergey Levine.
An Operator View of Policy Gradient Methods . NeurIPS 2020.
Dibya Ghosh, Marlos C Machado, Nicolas Le Roux
Representations for Stable Off-Policy Reinforcement Learning . ICML 2020.
Dibya Ghosh, Marc G. Bellemare.
On Catastrophic Interference in Atari 2600 Games Preprint.
William Fedus, Dibya Ghosh, John D. Martin, Marc G. Bellemare, Yoshua Bengio, Hugo Larochelle.
Learning Actionable Representations with Goal-Conditioned Policies . ICLR 2019.
Dibya Ghosh, Abhishek Gupta, Sergey Levine.
Variational Inverse Control with Events: A General Framework for Data-Driven Reward Definition . NeurIPS 2018.
Justin Fu, Avi Singh, Dibya Ghosh, Larry Yang, Sergey Levine.
Divide-and-Conquer Reinforcement Learning . ICLR 2018.
Dibya Ghosh, Avi Singh, Aravind Rajeswaran, Vikash Kumar, Sergey Levine.