BWhat changed
Trained an encoder to replace text embeddings with fewer continuous pooled representations and small decoder-specific mappings, supporting multiple unchanged decoders and compressed context extension.
Scientific scope
Trained an encoder to replace text embeddings with fewer continuous pooled representations and small decoder-specific mappings, supporting multiple unchanged decoders and compressed context extension. Lossy task-oriented compression requires encoder training and preprocessing; it is not lossless text storage or free context extension. Inherits soft compression and pretrained decoders. Portability is demonstrated for studied decoder families, not arbitrary models.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Lossy task-oriented compression requires encoder training and preprocessing; it is not lossless text storage or free context extension. Inherits soft compression and pretrained decoders. Portability is demonstrated for studied decoder families, not arbitrary models.
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.
- ARC-Encoder: learning compressed text representations for large language models
- Kyutai scientific publications index
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