DistilBERT: general pretraining distillation with a triple loss

What changed

Distilled BERT during pretraining with language-modeling, teacher-distillation and cosine-distance losses, yielding a reusable smaller representation model.

Scientific scope

Distilled BERT during pretraining with language-modeling, teacher-distillation and cosine-distance losses, yielding a reusable smaller representation model. Inherits BERT and knowledge distillation. Reported retained capability is a benchmark aggregate, not97percent of all possible language abilities; compact inference still requires the teacher/pretraining costs.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Inherits BERT and knowledge distillation. Reported retained capability is a benchmark aggregate, not97percent of all possible language abilities; compact inference still requires the teacher/pretraining costs.

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.

• DistilBERT, a distilled version of BERT: smaller, faster, cheaper and lighter

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