VLAFlow: A Unified Training Framework for Vision-Language-Action Models via Co-training and Future Latent Alignment

3 July 2026 at 06:00 CEST