🚀 Our method, Poutine, was the best-performing entry in the 2025 Waymo Vision-based End-to-End Driving Challenge at #CVPR2025! Our 3 B-parameter VLM Poutine scored 7.99 RFS on the official test set—comfortably ahead of every other entry (see figure).

Waymo is coming to London next year.

Waymos are 80% less likely to get into a serious crash than human drivers. An 80% reduction in car crash deaths in the US would mean more lives saved than if you eliminated all homicides. Great piece by @KelseyTuoc

“GRPO” is just rebranded REINFORCE. Everything “unique” about GRPO like the advantage normalization and (biased) KL regularization are pretty much useless. Kill GRPO. It’s always been REINFORCE.

NO verifiers. NO Tools. Qwen3-4B-Instruct can match DeepSeek-R1 and o3-mini (high) with ONLY test-time scaling. Presenting Recursive Self-Aggregation (RSA) — the strongest test-time scaling method I know of! Then we use aggregation-aware RL to push further!! 📈📈 🧵below!

🚨Announcing the World Modeling Workshop 2026 🚨 📅 When: Feb 4–6, 2026 📍Where: Mila (Montréal) + Online (free) 💡 What: Keynotes, Methods Deep Dive, and Tutorials 🌐 https://t.co/WukFtNON3o ✉️ worldmodel.mila@gmail.com 🧵 Details below:

@jxmnop Origin of most of these innovations is Canada 🇨🇦 though 😜

🚨 Excited to share our new work: "Stable Gradients for Stable Learning at Scale in Deep Reinforcement Learning"! 📈 We propose gradient interventions that enable stable, scalable learning, achieving significant performance gains across agents and environments! Details below 👇

How can we make behavioural cloning (BC) achieve better combinatorial generalization on out-of-distribution goals? We propose BYOL-γ: an auxiliary self-predictive loss to improve generalization for goal-conditioned BC. 🧵1/6

Poutine: Vision-Language-Trajectory Pre-Training and Reinforcement Learning Post-Training Enable Robust End-to-End Autonomous Driving.

Why do we keep sampling from the same distribution the model was trained on? We rethink this old paradigm by introducing Feynman-Kac Correctors (FKCs) – a flexible framework for controlling the distribution of samples at inference time in diffusion models! Without re-training

One of the most striking, non-text AI plots I've seen since ChatGPT launched. Scaling keeps working, this time for Waymo's tooling.

This was joint work with my amazing colleagues at @Mila_Quebec: Rodrigue de Schaetzen, @rogg1111 , @chrisjpal , @duckietown_coo Check out our report here:

Why did Poutine work?  • Plug-and-play VLM – Built on Qwen 2.5 VL 3B. No custom perception backbone or action headers needed.  • Simple and effective training recipe – Self-supervised vision-language-trajectory pre-training followed by lightweight RL preference fine-tuning.

This challenge pushed the limits of vision-based end-to-end planning in rare, long-tail scenarios. We show that VLMs can be repurposed into effective planners in the long-tail.

A residency clause made Québec teams ineligible for prizes, so we couldn’t collect the first-prize—but the challenge organizers awarded us a Special Mention instead. Thanks to the @Waymo challenge organizers for the shout-out!

Excited that our paper "Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization" was accepted to ICML 2025! We show how Preference Optimization can reduce the impact of noisy concept labels in CBMs. 🧵/9

(1/n)🚨You can train a model solving DFT for any geometry almost without training data!🚨 Introducing Self-Refining Training for Amortized Density Functional Theory — a variational framework for learning a DFT solver that predicts the ground-state solutions for different

New preprint! 🧠🤖 How do we build neural decoders that are: ⚡️ fast enough for real-time use 🎯 accurate across diverse tasks 🌍 generalizable to new sessions, subjects, and species? We present POSSM, a hybrid SSM architecture that optimizes for all three of these axes! 🧵1/7

🚗💥Introducing Ctrl-Crash: controllable video generation for autonomous driving! SOTA models struggle to generate physically realistic car crashes. We propose an image2video diffusion model with bounding box and crash type control. Website: https://t.co/vNBYhbx3c4 🧵->