Yang and colleagues: XLNet permutation language-model pre-training

What changed

Used permutations of autoregressive factorization and a two-stream attention design to learn bidirectional language representations.

Scientific scope

Used permutations of autoregressive factorization and a two-stream attention design to learn bidirectional language representations. Factorization order is permuted, not necessarily the input token positions. Transformer-XL recurrence is inherited; benchmark gains are not universal dominance.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Factorization order is permuted, not necessarily the input token positions. Transformer-XL recurrence is inherited; benchmark gains are not universal dominance.

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.

• XLNet: Generalized Autoregressive Pretraining for Language Understanding

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