Sergey Levine (@svlevine) / X

Sergey Levine

2,590 posts

Sergey Levine

@svlevine

Associate Professor at UC Berkeley Co-founder, Physical Intelligence

Berkeley, CA

rail.eecs.berkeley.edu

Joined April 2018

Sergey Levine
@svlevine
Mar 13, 2021
I'm releasing all the lectures (so far) for my deep learning class, CS182! This is an introductory deep learning course (advanced undergraduate + graduate) covering a broad range of deep learning topics. Website: cs182sp21.github.io Playlist: youtube.com/playlist?list=… 🧵->
Sergey Levine
@svlevine
Oct 11, 2020
My deep RL course (CS285) now has fall 2020 lectures online, here: youtube.com/playlist?list=… We'll update this each week with the latest lectures. Hopefully these lectures are helpful! We tried to update material from past years, and recorded it in a more online-friendly format.
youtube.com
Deep Reinforcement Learning: CS 285 Fall 2020
Lectures for UC Berkeley CS 285: Deep Reinforcement Learning.
Sergey Levine
@svlevine
Sep 29, 2019
Want to learn deep RL? My deep RL course now has a permanent course number (CS285) and is being offered this semester: rail.eecs.berkeley.edu/deeprlcourse/ Lecture videos here (so far, we've gotten through most of model-free RL, model-based RL coming up next): youtube.com/playlist?list=…
Sergey Levine
@svlevine
Apr 18, 2021
Final set of CS182 Deep Learning lectures now added to the course playlist: youtube.com/playlist?list=… GANs (Lec 19), adv. examples (20), and meta-learning (21)! More materials on the course website: cs182sp21.github.io
youtube.com
Deep Learning: CS 182 Spring 2021
Lectures for UC Berkeley CS 182: Deep Learning.
Sergey Levine
@svlevine
Jun 8, 2025
I always found it puzzling how language models learn so much from next-token prediction, while video models learn so little from next frame prediction. Maybe it's because LLMs are actually brain scanners in disguise. Idle musings in my new blog post:
Language Models in Plato's Cave
From sergeylevine.substack.com
315K
Sergey Levine
@svlevine
Mar 12, 2024
Since cat is out of the bag, it’s time I share: I’ll be starting a new adventure with an incredible team of friends and long-time collaborators to take on the big challenge of robot learning at scale! It's called Physical Intelligence (Pi… or π, like the symbol). 🧵👇
109K
Sergey Levine
@svlevine
Apr 5, 2021
I've updated the CS182: Deep Learning with new lectures on RL (policy gradients, actor critic, Q-learning), autoencoders, latent variable models, and VAEs! Website: cs182sp21.github.io Playlist (lectures 15-18): youtube.com/playlist?list=… GANs and adv examples coming next!
Sergey Levine
@svlevine
Jun 26, 2025
If you have a policy that uses diffusion/flow (e.g. diffusion VLA), you can run RL where the actor chooses the noise, which is then denoised by the policy to produce an action. This method, which we call diffusion steering (DSRL), leads to a remarkably efficient RL method! 🧵👇
00:00
153K
Sergey Levine
@svlevine
Oct 31, 2024
Really excited to share what I've been working on with my colleagues at Physical Intelligence! We've developed a prototype robotic foundation model that can fold laundry, assemble a box, bus a table, and many other things. We've written a paper and blog post about it. 🧵👇
00:00
115K
Sergey Levine
@svlevine
Feb 8, 2021
What did we learn from 5 years of robotic deep RL? My colleagues at Google and I tried to distill our experience into a review-style journal paper, covering some of the practical aspects of real-world robotic deep RL: arxiv.org/abs/2102.02915 🧵->
Sergey Levine
@svlevine
Jan 11, 2023
We're releasing our code for "driving any robot", so you can also try driving your robot using the general navigation model (GNM): github.com/PrieureDeSion/… Code goes with the GNM paper: sites.google.com/view/drive-any… Should work for locobot, hopefully convenient to hook up to any robot
GIF
113K
Sergey Levine
@svlevine
May 22, 2023
We figured out how to train diffusion models with RL to generate images aligned with user goals! Our RL method gets ants to play chess and dolphins to ride bikes. Reward from powerful vision-language models (i.e., RL from AI feedback): rl-diffusion.github.io A 🧵👇
130K
Sergey Levine
@svlevine
Feb 4, 2025
Very happy to announce that we are open-sourcing the π₀ model, weights, and some fine-tuned checkpoints! Hoping this leads to lots of great follow-up research: github.com/Physical-Intel… Here is a fun test from our friends at UPenn.
00:00
72K
Sergey Levine
@svlevine
Dec 17, 2020
All of the main lectures for UC Berkeley's fall 2020 deep RL course are now posted: youtube.com/playlist?list=… Newly posted lectures: 21 (transfer learning), 22 (meta-learning), 23 (open problems)!