ALPS unifies language, reconstruction and state in one visual encoder
Frozen ALPS tokens reach 85.5% mean policy success on LIBERO and match a 3.5x larger V-JEPA 2-L on state probes.
4 stories, newest first · one card per story
Frozen ALPS tokens reach 85.5% mean policy success on LIBERO and match a 3.5x larger V-JEPA 2-L on state probes.
UnAct forgets classes with forward passes only and stays close to retraining when data is scarce.
TT-VidT pairs a DINOv3 spatial ViT with a temporal layer, leading motion benchmarks at lower FLOPs.
WeightPE finetunes int8 weights so Re-Pair grammars shrink to 0.38–0.43x of QAT, costing about 1–2 accuracy points.
Today 1.2K readers · 1.4K views · 193 reads · all time 203K views