David McAllister

PhD Student @berkeley_ai

[David McAllister @davidrmcall]

Excited to share Flow Matching Policy Gradients: expressive RL policies trained from rewards using flow matching. It’s an easy, drop-in replacement for Gaussian PPO on control tasks.

[David McAllister @davidrmcall]

@redstone_hong @arthurallshire

[David McAllister @davidrmcall]

@_sam_sinha_ @BerntBornich @1x_tech Congrats!

[David McAllister @davidrmcall]

Robot (@arthurallshire)

[David McAllister @davidrmcall]

@_sam_sinha_ @LumaLabsAI Congrats Sam !

[David McAllister @davidrmcall]

@ronskoro Thanks! We didn’t try with distilled students though that sounds promising. Our experiments initialized from pretrained SD3 checkpoints, would you consider that late training?

[David McAllister @davidrmcall]

We developed a simple, sample-efficient online RL technique for post-training image generation models. We see it as a possible steerable alternative to CFG, driven by any scalar reward, including human preference.

[David McAllister @davidrmcall]

@ayaanzhaque Congrats Ayaan!

[David McAllister @davidrmcall]

Check out our blog post at https://t.co/W9CqU9yIlJ for a walkthrough of our design decisions.

w/ fantastic collaborators: Miika Aittala, Tero Karras, Janne Hellsten, @akanazawa Timo Aila and Samuli Laine

[David McAllister @davidrmcall]

@KushtimusPrime Congrats!

[Kush Hari @KushtimusPrime]

Our new work, STITCH 2.0, can perform consecutive running sutures to close a sample wound with the daVinci robot.

[David McAllister @davidrmcall]

@vonekels Congrats!

[Junyi Zhang @CVPR @junyi42]

𝗢𝗻𝗲 𝗺𝗲𝗺𝗼𝗿𝘆 𝗰𝗮𝗻’𝘁 𝗿𝘂𝗹𝗲 𝘁𝗵𝗲𝗺 𝗮𝗹𝗹.

We present 𝗟𝗼𝗚𝗲𝗥, a new 𝗵𝘆𝗯𝗿𝗶𝗱 𝗺𝗲𝗺𝗼𝗿𝘆 architecture for long-context geometric reconstruction.

LoGeR enables stable reconstruction over up to 𝟭𝟬𝗸 𝗳𝗿𝗮𝗺𝗲𝘀 / 𝗸𝗶𝗹𝗼𝗺𝗲𝘁𝗲𝗿 𝘀𝗰𝗮𝗹𝗲, with 𝗹𝗶𝗻𝗲𝗮𝗿-𝘁𝗶𝗺𝗲 𝘀𝗰𝗮𝗹𝗶𝗻𝗴 in sequence length, 𝗳𝘂𝗹𝗹𝘆 𝗳𝗲𝗲𝗱𝗳𝗼𝗿𝘄𝗮𝗿𝗱 inference, and 𝗻𝗼 𝗽𝗼𝘀𝘁-𝗼𝗽𝘁𝗶𝗺𝗶𝘇𝗮𝘁𝗶𝗼𝗻.

Yet it matches or surpasses strong optimization-based pipelines. (1/5)

@GoogleDeepMind @Berkeley_AI

[Angjoo Kanazawa @akanazawa]

@brenthyi who worked on FPO/FPO++ is finishing his PhD and going on the job market 😭✨

He is also the person behind viser, pyroki, egoallo, jaxls, tyro and more!

I can't express how amazing it is to have Brent on your team..! Any team would be incredibly lucky to have him!!

[David McAllister @davidrmcall]

@Songwei_Ge That's a sharp dog!

[David McAllister @davidrmcall]

@AnthropicAI @justkerrding they want your tokens

[Grace Luo @graceluo_]

We trained diffusion models on a billion LLM activations, and we want you to use them!

New preprint: Learning a Generative Meta-Model of LLM Activations
Joint work with @feng_jiahai, @trevordarrell, @AlecRad, @JacobSteinhardt.

More in thread 🧵

[David McAllister @davidrmcall]

Brent and Hongsuk improved FPO to control robots with flow/diffusion models in real time!

[Brent Yi @brenthyi]

New project! Flow Policy Gradients for Robot Control

tldr; a simple online RL recipe for training and fine-tuning flow policies for robots

co-led w/ @redstone_hong: https://t.co/nKSq9EakUy

[Fanqi Lin @lfqirrrrr]

𝑪𝒐-𝒕𝒓𝒂𝒊𝒏𝒊𝒏𝒈 is a promising way to scale Large Behavior Models (LBMs) beyond robot data, yet the data and training recipe are far from settled. 🤔

We present a large-scale empirical study leveraging 4,000h of robot/human data and 50M vision-language samples, evaluating 89 policies across 58,000 simulation rollouts and 2,835 real-world trials. 🤖📊

https://t.co/jMlZWdXexl
Work done during my internship at @ToyotaResearch.

[Brent Yi @brenthyi]

tyro 1.0 is out 🐣

This has been a pet project/niche interest of mine for ~4 years now, so it's a bit of a sentimental moment...

https://t.co/bAibP3RjxE

[Qiyang (Colin) Li @qiyang_li]

Action chunking is drawing growing interest in RL, yet its theoretical properties are still understudied.

We are excited to share some insights on when we should use action chunking in Q-learning + a new algo (DQC) to tackle hard long-horizon tasks!https://t.co/izVWQBgH3c🧵1/N