Encoders · 2020

ELECTRA: Pre-training Text Encoders as Discriminators

Kevin Clark, Minh-Thang Luong, Quoc V. Le, Christopher D. Manning

ELECTRA trains an encoder with replaced-token detection, a discriminator that classifies every token as original or generator-substituted, extracting learning signal from all positions instead of only the ~15% masked ones and cutting pretraining compute.

Editorial record

Plain-language summary

Rather than masking tokens and predicting them, ELECTRA uses a small generator to swap some tokens for plausible alternatives, then trains the main model to decide, for every token, whether it was replaced. Because the loss covers all positions rather than the small masked subset, each training step yields more signal per example. This let ELECTRA match or exceed BERT-scale accuracy using substantially less pretraining compute, making strong encoders reachable on smaller budgets.

Knowledge graph

Relationships

Descendants

Source record

Provenance

Record ID
P-016
Record created
2026-07-13
Last reviewed
2026-07-14
Record version
2

Citation caveat: Citation metadata is approximate and marked unverified in the source dataset.