BWhat changed
Encoded moderately masked text into a sentence vector and reconstructed from a more heavily masked input with a lightweight decoder, adding position-specific decoding for denser training signals.
Scientific scope
Encoded moderately masked text into a sentence vector and reconstructed from a more heavily masked input with a lightweight decoder, adding position-specific decoding for denser training signals. Reconstruction also uses visible token context, not the sentence vector alone; it is not a lossless text-compression guarantee. Masked autoencoding, two-stream/position masks and pretrained Transformers are inherited. Original affiliations include Huawei and BUPT, with a prior-Microsoft work note, not retrospective BAAI authorship.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Reconstruction also uses visible token context, not the sentence vector alone; it is not a lossless text-compression guarantee. Masked autoencoding, two-stream/position masks and pretrained Transformers are inherited. Original affiliations include Huawei and BUPT, with a prior-Microsoft work note, not retrospective BAAI authorship.
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.
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