Signal2026-08-10
arXiv

TEMPO: Semantic-Action Decoupled RL Post-Training for Vision-Language-Action Models

Part of

Real-Time Action Conditioned Video Generation

Open primary sourceSee the whole picture