Lan and colleagues: ALBERT parameter-efficient language representations

What changed

Reduced BERT-style model parameters through factorized embeddings and cross-layer sharing while evaluating language-understanding performance.

Scientific scope

Reduced BERT-style model parameters through factorized embeddings and cross-layer sharing while evaluating language-understanding performance. Scope limitation: First preprint appeared in 2019; year follows ICLR 2020. ALBERT is a BERT-derived design, not original bidirectional pre-training.

Evidence

The retained primary-source abstract, bibliographic record and stated review scope support this provisional assessment. No independent experimental replication by WRS is claimed. First preprint appeared in 2019; year follows ICLR 2020. ALBERT is a BERT-derived design, not original bidirectional pre-training.

Contributors

• The named authors listed in the linked research record · SCIENCE: Complete source byline retained in the research record; human research direction, method development and evaluation. Individual labor shares are not measured.

• Lan and colleagues: ALBERT parameter-efficient language representations — original paper
• Lan and colleagues: ALBERT parameter-efficient language representations — official company source

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