BWhat changed
Exposed a compact task-relevant token from a pretrained VLA and trained lightweight actor/critic heads with a regularizer anchoring actions to the frozen VLA, enabling online adaptation of precise manipulation.
Scientific scope
Exposed a compact task-relevant token from a pretrained VLA and trained lightweight actor/critic heads with a regularizer anchoring actions to the frozen VLA, enabling online adaptation of precise manipulation. Actor–critic, residual control, frozen representations and VLA pretraining are inherited. Acquired delta is the learned compact interface/anchored adaptation recipe, not onlineRL itself. Four task improvements do not imply arbitrary-task sample efficiency or physical safety.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Actor–critic, residual control, frozen representations and VLA pretraining are inherited. Acquired delta is the learned compact interface/anchored adaptation recipe, not onlineRL itself. Four task improvements do not imply arbitrary-task sample efficiency or physical safety.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- Xu et al.: compact frozen-VLA interface for online actor–critic adaptation — original technical report
- Xu et al.: compact frozen-VLA interface for online actor–critic adaptation — institutional account
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction