← back to paper
arxiv: 2605.10819 · 2 revisions
ALAM: Algebraically Consistent Latent Action Model for Vision-Language-Action Models