Luke Rowe

🚀 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).

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. arxiv.org/abs/2506.11234

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.

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: anthonygosselin.github.io/Ctrl-Crash-Pro… 🧵->

Is AdamW the best inner optimizer for DiLoCo? Does the inner optimizer affect the compressibility of the DiLoCo delta? Excited to introduce MuLoCo: Muon is a practical inner optimizer for DiLoCo! 🧵arxiv.org/abs/2505.23725 1/N

Confused about recent LLM RL results where models improve without any ground-truth signal? We were too. Until we looked at the reported numbers of the Pre-RL models and realized they were serverely underreported across papers. We compiled discrepancies in a blog below🧵👇

Thanks @_akhaliq for sharing our work! Excited to present our next generation of SVG models, now using Reinforcement Learning from Rendering Feedback (RLRF). 🧠 We think we cracked SVG generalization with this one. Go read the paper! arxiv.org/abs/2505.20793 More details on…

Is there a universal strategy to turn any generative model—GANs, VAEs, diffusion models, or flows—into a conditional sampler, or finetuned to optimize a reward function? Yes! Outsourced Diffusion Sampling (ODS) accepted to @icmlconf , does exactly that!

We find training unified multimodal understanding and generation models is so easy, you do not need to tune MLLMs at all. MLLM's knowledge/reasoning/in-context learning can be transferred from multimodal understanding (text output) to generation (pixel output) even it is FROZEN!