François Porcher

PhDing @Meta and @Inria | SAM3, World Modeling, Robotics | @UCBerkeley

[François Porcher @foporcher]

1/ My first PhD paper is out! 🎓

Title: Flow Matching in Feature Space for Stochastic World Modeling

tldr: we build stochastic world models directly in high-dimensional DINOv3 feature space, instead of relying on low-dimensional VAE latents.

[François Porcher @foporcher]

@omp87 @CSProfKGD Thanks!!

[François Porcher @foporcher]

1/ My first PhD paper is out! 🎓

Title: Flow Matching in Feature Space for Stochastic World Modeling

tldr: we build stochastic world models directly in high-dimensional DINOv3 feature space, instead of relying on low-dimensional VAE latents.

[François Porcher @foporcher]

@Em_Nomadic Thank you!

[François Porcher @foporcher]

@tahboub_h Thank you!

[François Porcher @foporcher]

Pretty cool work which does planning in the policy space instead of action space, and with jumpy world models!

[Ahmed Touati @ahtouati]

How do we get RL agents to plan over long horizons without drowning in compounding errors?

Instead of reasoning over primitive, step-by-step actions, what if they could compose pre-trained policies across multiple timescales?

Poster #1117 Wed 8th at 2:30 pm #ICML2026

[François Porcher @foporcher]

10/ Huge thanks to my co-authors Nicolas Carion (@alcinos26), Karteek Alahari (@inthebrownbag), and Shizhe Chen 🙏

Paper: https://t.co/9CTt2Fa1Yu
Code: https://t.co/9pC3SoL2os

[François Porcher @foporcher]

9/ Results 📊

FlowWM improves World Modeling for perception tasks, mode coverage, and horizon robustness.

We test this both on FuturePerception and on Bouncing Shapes, where we can enumerate all possible stochastic futures and measure precision/recall/F1.

[Jarod Levy @JarodLevy]

💃 DANCE 🕺: Detect and Classify Events in EEG

Today, we’re releasing DANCE, an end-to-end model that detects and classifies events in EEG in one pass.

📄 Paper: https://t.co/0TLhzxXHK0
💻 Code: https://t.co/NEfQ7wxXLJ

🧵🧵🧵 More details below 🧵🧵🧵

[Joséphine Raugel @JRaugel]

1/ We’re so glad to share this new study 💫
Does the brain learn like a Deep Net? 🧠⚙️
- 📄Misalignment Between Backpropagation and the Hierarchy of Brain Responses to Images
- 🔗https://t.co/yrb4otBEYk
Thread below 🧵

[Kunhao Zheng @KunhaoZ]

🧵 For 2 RL checkpoints trained differently, you can just weight extrapolate them and it works!

Bonus: these extrapolated checkpoints are complementary policies
-> Get exploration and diversity for free
-> Better inference scaling when ensembling

Paper: https://t.co/zU0LH0TOdm

[François Porcher @foporcher]

@arnal_charles @KempeLab Congrats Charles!!

[Charles Arnal @arnal_charles]

(1/9) Experience replay can cut LLM RL training compute by up to ~40% (without hurting final accuracy—and sometimes improving it).

Paper: https://t.co/6YcAd6EBSy

[François Porcher @foporcher]

@mathemetica Why is the initial noise distribution not fully gaussian?

[Fabian Gloeckle @FabianGloeckle]

A new milestone in automatic formalization:
We translated an entire graduate math textbook into Lean using 30K LLM agents.
Open-source, large-scale multi-agent inference that actually works
> Blueprint+Lean: https://t.co/YkhG2g1cWp
> Codebase+preprint: https://t.co/EH3IphOKZ3
1/7