Alexandre Brown 🇨🇦 (@AlexandreBrown0) header
Alexandre Brown 🇨🇦 (@AlexandreBrown0) profile picture

PhD student, researcher at @Mila_Quebec ~ Working on humanoids & RL

Canada
Joined September 2017
Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)Photo by Alexandre Brown 🇨🇦 (@AlexandreBrown0)
Alexandre Brown 🇨🇦 retweeted
(1/4) why do image models chop every picture into the same square patches? 🤔 (hint: they shouldn't) following the bitter lesson, we let diffusion models *choose* how to tokenize an image. it spends fewer tokens on the boring background, and more tokens on our faces! 🧵
74
144
28
2,678
148,198
I was curious how much the style reward mattered in BFM-Zero. So I removed it from the actor loss. The policy got significantly worse... and a lot more like C3PO 🤖, which I found quite amusing 😅. Turns out style guidance matters more than I thought, even for steering the Forward Backward representations.
2
2
28
1,429
Motion tracking seems to hold up better without actor style guidance than goal reaching or zero-shot reward inference. Still looks and performs slightly worse than with style guidance. Perhaps the constant C3PO stomping is just less awkward when the target is moving too!
1
1
53
PS: C3PO is still cool tho 😎
1
47
Alexandre Brown 🇨🇦 retweeted
ChatGPT required RL, it was not pure imitation learning or dagger; reasoning models also required RL. Obviously the same will be true of robotics, the question is just if base models are good enough
pendulum has fully swung back from where it was last year (“scale up data collection, behavior clone large models”). everyone is finally seeing the shining light of RL!
6
6
93
4,401
Reproduced BFM-Zero results in sim. At 65M train steps, about 1/3 of the released checkpoint's 211M. The policy already looks promising on these goal reaching examples. The released checkpoint appears stronger on some harder poses but the qualitative gap is smaller than I expected. Zero-shot unsupervised RL looks promising 🔥!
10
4
51
1,488
More lookahead barely changed zero-shot motion tracking in my tests. Averaging 1, 2, 4, 6, 8, 10, or 12 reference frames produced visually similar results for both checkpoints. One possible reason: B receives reference velocities, so even a one-frame prompt (z) could encode motion information.
1
1
132
Zero-shot reward inference with BFM-Zero worked surprisingly well 👌. My 65M steps vs the author's 211M. The flow is pretty simple: Define a reward function → infer z → act. No fine-tuning. The policy was never trained on any of these reward functions which is super cool IMO. I observed strong behaviors overall, with room for more precise speeds and poses. Random replay samples were used to compute z. No sample or prompt cherry-picking.
1
73
Alexandre Brown 🇨🇦 retweeted
GPT-6 Astra's robotic capabilities are not emergent. It was trained on robotics specific data. So were GPT-6.1 Sol and Gemini Argon Here's how we know 🧵
34
78
26
1,066
133,291
Alexandre Brown 🇨🇦 retweeted
Replying to @theSamPadilla
It's the most open they have been in years ;).
1
3
343
Alexandre Brown 🇨🇦 retweeted
i kinda miss manually debugging things
36
38
8
390
13,945
Alexandre Brown 🇨🇦 retweeted
apple trained a unitree g1 humanoid only in simulation, across 1000 randomly generated houses then it went into 6 real homes it had never seen, no fine tuning, no scans of the rooms 360 trials, 91% success crouch under a table: 100%. go through a door: 93%. sit down: 85% very bullish on sim
2
1
29
1,777
Alexandre Brown 🇨🇦 retweeted
the more i code with LLMs, the more i feel the urge to go back to fundamentals and understand things as deeply as possible.
41
88
10
1,254
22,024
Had some fun today trying out GPT 6.1 Terra + SONIC on a Unitree G1. Zero-shot and fully autonomous using only the egocentric camera of the G1 and proprioception! Task: reach the red balloon 🎈 Three successful reaches. Zero lab escapes 😆
1
6
676