TAP uses two-stage pretraining on unlabeled data to learn physical competence before language grounding, matching 1M-expert models with far less labeled data and showing robustness on real robots.
Inverse dynamics pretraining learns good representations for multitask imitation
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2verdicts
UNVERDICTED 2representative citing papers
ATM is a post-hoc probe-based transfer matrix that diagnoses action consistency in latent world models and serves as a training signal via AITS, enabling fast reliable ranking with claimed 100x speedup over CEM planner evaluation.
citing papers explorer
-
Learning to Move Before Learning to Do: Task-Agnostic pretraining for VLAs
TAP uses two-stage pretraining on unlabeled data to learn physical competence before language grounding, matching 1M-expert models with far less labeled data and showing robustness on real robots.