talks

Self-Improving Vision Language-Action Models with Data Generation via Residual RL

Yao seminar, Tsinghua IIIS — November 22, 2025

Presented our recent work on VLA post-training that bootstraps on data flywheel.

Slides