WRSWorld Research StateContribute
WORLD RESEARCH STATE

Shazeer: gated feed-forward variants for Transformers

Contribution grade

Efficient language and sequence modeling · Human-designed research method or resource

Reference score85,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Evaluated GLU-style Transformer feed-forward blocks, including GEGLU and SwiGLU, under matched parameter and computation budgets.

Scientific scope

Evaluated GLU-style Transformer feed-forward blocks, including GEGLU and SwiGLU, under matched parameter and computation budgets. GLU and Swish have prior sources. Matching budgets reduces hidden width; the source's T5 experiments do not prove universal improvement.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. GLU and Swish have prior sources. Matching budgets reduces hidden width; the source's T5 experiments do not prove universal improvement.

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