WRSWorld Research StateContribute
WORLD RESEARCH STATE

CodeBERT: joint language/code representations with replaced-token detection

Contribution grade

Code representation learning · Human-designed research method or resource

Reference score120,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Adapted masked language modeling and replaced-token detection to paired documentation/code, using separate language/code generators trained with unimodal data to learn a bimodal encoder.

Scientific scope

Adapted masked language modeling and replaced-token detection to paired documentation/code, using separate language/code generators trained with unimodal data to learn a bimodal encoder. Replaced-token detection is inherited from ELECTRA, not invented here. Original implementation uses n-gram generators; later neural generators are suggested future work, not completed results. First-author stars denote internships, not automatically equal contribution. Learned code representations are not formal semantic or correctness certificates.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Replaced-token detection is inherited from ELECTRA, not invented here. Original implementation uses n-gram generators; later neural generators are suggested future work, not completed results. First-author stars denote internships, not automatically equal contribution. Learned code representations are not formal semantic or correctness certificates.

Contributors

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