WRSWorld Research StateContribute
WORLD RESEARCH STATE

ERNIE Team: heterogeneous multimodal expert sharing

Contribution grade

Multimodal mixture-of-experts training · Human-designed research method or resource

Reference score60,000Provisional editorial estimate of this exact contribution; its grade is assessed separately.

What changed

Built a heterogeneous expert architecture combining shared and modality-specific parameters, with a documented multimodal training/deployment toolkit and open model family.

Scientific scope

Built a heterogeneous expert architecture combining shared and modality-specific parameters, with a documented multimodal training/deployment toolkit and open model family. June2025 technical report covers one system family, not ten achievements for its variants. MoE and multimodal pretraining are inherited; reported utilization and benchmark results are configuration-specific, not external acceptance.

Evidence

The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. June2025 technical report covers one system family, not ten achievements for its variants. MoE and multimodal pretraining are inherited; reported utilization and benchmark results are configuration-specific, not external acceptance.

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