Representation also matters for VLA models! Introducing FLARE: Robot Learning with Implicit World Modeling. With future latent alignment objective, FLARE significantly improves policy performance on multitask imitation learning & unlocks learning from egocentric human videos.

Speed up your diffusion models—with evolutionary caching! Proud to share Ani Aggarwal’s first paper (mentored by Matt Gwilliam)! ECAD uses evolutionary search to find Pareto-optimal schedules that speed up diffusion & boost quality—no retraining or tuning. Read the full 🧵below

🎉 Excited to share our paper "Trokens: Semantic-Aware Relational Trajectory Tokens for Few-Shot Action Recognition" was accepted to #ICCV2025! Co-led with Pulkit in collaboration with Matthew Walmer, Saketh Rambhatla , and Abhinav Shrivastava Website:trokens-iccv25.github.io