LayoutMask: Enhance Text-Layout Interaction in Multi-modal Pre-training for Document Understanding

Benchmark Model Rank Results
key-information-extraction-on-cordLayoutMask (base)–F1: 96.99
key-information-extraction-on-cordLayoutMask (large)–F1: 97.19
named-entity-recognition-ner-on-cord-rLayoutMask–F1: 81.84
named-entity-recognition-ner-on-funsd-rLayoutMask–F1: 77.10
semantic-entity-labeling-on-funsdLayoutMask (base)–F1: 92.91
semantic-entity-labeling-on-funsdLayoutMask (large)–F1: 93.20