1. X
  2. Kevin Black
Log inSign up
Kevin Black
133 posts
user avatar
Kevin Black
@kvablack
phd @berkeley_ai, research @physical_int
Joined March 2018
133
Following
3,295
Followers
RepliesRepliesMediaMedia

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • user avatar
    Kevin Black
    @kvablack
    Oct 31, 2024
    It's been 6 months since I slammed the brakes on several PhD research projects to go work at π... 😅 super excited to finally share our results! A short 🧵 with some details:
    user avatar
    Physical Intelligence
    @physical_int
    Oct 31, 2024
    At Physical Intelligence (π) our mission is to bring general-purpose AI into the physical world. We're excited to show the first step towards this mission - our first generalist model π₀ 🧠 🤖 Paper, blog, uncut videos: physicalintelligence.company/blog/pi0
    Image
    00:00
    102K
  • user avatar
    Kevin Black
    @kvablack
    Jun 9, 2025
    In LLM land, a slow model is annoying. In robotics, a slow model can be disastrous! Visible pauses at best, dangerously jerky motions at worst. But large VLAs are slow by nature. What can we do about this? An in-depth 🧵:
    Image
    00:00
    90K
  • user avatar
    Kevin Black
    @kvablack
    Jul 5, 2023
    The biggest problem with our RL diffusion paper was that nobody could run our Jax+TPU code. No more! I've reimplemented DDPO in PyTorch, plus replicated our results using LoRA for low-memory training! Links below 👇
    Image
    00:00
    91K
  • user avatar
    Kevin Black
    @kvablack
    Sep 24, 2024
    These people copy + pasted the website I designed (including the body text), changed some stuff here and there, and then released it as their own work without attribution. The best part is they didn't even change the preview thumbnail so it still says "Octo". Incredible stuff!
    user avatar
    Chris Paxton
    @chris_j_paxton
    Sep 23, 2024
    Replying to @chris_j_paxton
    TinyVLA paper: arxiv.org/pdf/2409.12514 project page: tiny-vla.github.io
    74K
  • user avatar
    Kevin Black
    @kvablack
    Nov 12, 2024
    Here's a link to the recording for anyone that's interested! youtube.com/live/ELUMFpJCU…
    user avatar
    Kevin Black
    @kvablack
    Nov 7, 2024
    If you're at #CoRL2024, come check out my talk at the X-Embodiment workshop at 1:30pm! Thanks to @KarlPertsch for inviting me to speak!
    From Octo to π0: How to Train Your Generalist Robot Policy
    20K
  • user avatar
    Kevin Black
    @kvablack
    Nov 7, 2024
    If you're at #CoRL2024, come check out my talk at the X-Embodiment workshop at 1:30pm! Thanks to @KarlPertsch for inviting me to speak!
    From Octo to π0: How to Train Your Generalist Robot Policy
    39K
  • user avatar
    Kevin Black
    @kvablack
    Jun 9, 2025
    This caption is a bit funny to me because we've put precisely zero effort into optimizing our model implementation. Thanks JAX!
    "Also, despite its larger size, π0 outperforms both RDT-1B and Diffusion Policy in speed thanks to its optimized JAX implementation (all other methods are implemented in PyTorch)."
    14K
  • user avatar
    Kevin Black
    @kvablack
    Nov 12, 2024
    My favorite slide that I made for my talk last weekend -- a very silly thought experiment in which we compare language datasets to robotics datasets (in the most shallow way possible). Yes it is to scale; I learned that the maximum shape size in Keynote is 20,000pts
    Comparison of robotics and language datasets in terms of hours: OXE, π dataset, GPT-2 training dataset, and Llama 3 training dataset.
    21K
  • user avatar
    Kevin Black
    @kvablack
    Jul 12, 2024
    I'm surprised more window manager enthusiasts don't know about yabai. This is what my macOS setup looks like -- maybe I'm just not a power user, but I don't miss i3 one bit!
    Image
    00:00
    user avatar
    Lucas Beyer (bl16)
    @giffmana
    Jul 11, 2024
    Replying to @enriquezaf_ and @awesome_ruler_
    No you cannot, you can only get it down to 200ms or so. At least when i had one 4y ago, and believe me i searched!
    22K
  • user avatar
    Kevin Black
    @kvablack
    Nov 28, 2024
    Replying to @cloneofsimo
    not just RoPE, tons of ppl copy the original Vaswani et. al. posemb code without thinking. it's *much much worse* for diffusion/flow timestep encodings; if you're not careful, you can end up using an encoding calibrated for t ∈ [0, 1000] with t ∈ [0, 1].
    37K
  • user avatar
    Kevin Black
    @kvablack
    Aug 28, 2023
    🔥🔥🔥 wake up babe, new BridgeData just dropped 🔥🔥🔥 Are you a fan of the original BridgeData? Doesn't matter! BridgeData V2 has 60k trajectories, 24 environments, 13 skills, and 100+ objects.
    Image
    00:00
    13K
  • user avatar
    Kevin Black
    @kvablack
    Feb 23, 2025
    Replying to @kevin_zakka
    bro capitalizes TeleOperate like it's a Special Move from a fantasy novel
    3.3K
  • user avatar
    Kevin Black
    @kvablack
    Dec 14, 2023
    We combined the 3 hottest things in machine learning: transformers, diffusion, and cute animal names, and what we got was Octo🐙: an open-source, cross-embodied, generalist robot policy backbone! 1/n
    Image
    00:00
    8.9K
  • user avatar
    Kevin Black
    @kvablack
    May 8, 2024
    If you're at #ICLR2024, come check out my DDPO poster tomorrow -- Thurs 10:45am, poster #21, Hall B! It's crazy looking back to see how much has changed since the paper first came out nearly 1 year ago. Really makes me feel how fast things move in this field.
    Image
    Image
    00:11
    user avatar
    Sergey Levine
    @svlevine
    Oct 1, 2023
    We've updated the DDPO website with some new results for training diffusion models with RL! Our aesthetic bunny is now much more... aesthetic. Latest here: rl-diffusion.github.io Includes code, LoRA training for low memory, pretrained models, etc. Some highlights 👇
    10K
Advertisement
Advertisement