AWhat changed
Compressed continuous action chunks through discrete-cosine-transform-based tokenization before autoregressive learning, releasing a tokenizer trained across diverse robot embodiments and control rates.
Scientific scope
Compressed continuous action chunks through discrete-cosine-transform-based tokenization before autoregressive learning, releasing a tokenizer trained across diverse robot embodiments and control rates. DCT, compression and subword/token modeling are inherited. Acquired delta is action-sequence encoding for VLA training, not invention of DCT or a universal tokenizer guarantee. Training-speed comparisons are workload-specific; FAST+ and pi0-FAST bundled.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. DCT, compression and subword/token modeling are inherited. Acquired delta is action-sequence encoding for VLA training, not invention of DCT or a universal tokenizer guarantee. Training-speed comparisons are workload-specific; FAST+ and pi0-FAST bundled.
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.
- FAST: Efficient Action Tokenization for Vision-Language-Action Models
- Pertsch et al.: frequency-space robot-action tokenization — 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