A policy model trained jointly with JEPA-style latent observation-sequence prediction matches ACT baselines in Meta-World and improves a proprioceptive reconstruction probe, though the task-success gain is within error bars.
Videomae: Masked autoencoders are data-efficient learners for self-supervised video pre-training , volume =
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
ACT-JEPA: Novel Joint-Embedding Predictive Architecture for Efficient Policy Representation Learning
A policy model trained jointly with JEPA-style latent observation-sequence prediction matches ACT baselines in Meta-World and improves a proprioceptive reconstruction probe, though the task-success gain is within error bars.