OpenCodePapers

semantic-entity-labeling-on-funsd

Semantic entity labeling
Dataset Link
Results over time
Click legend items to toggle metrics. Hover points for model names.
Leaderboard
Click a sortable column header to sort.
PaperCodeF1ModelNameReleaseDate
LayoutMask: Enhance Text-Layout Interaction in Multi-modal Pre-training for Document Understanding93.20LayoutMask (large)2023-05-30
ERNIE-Layout: Layout Knowledge Enhanced Pre-training for Visually-rich Document Understanding✓ Link93.12ERNIE-Layoutlarge2022-10-12
LayoutMask: Enhance Text-Layout Interaction in Multi-modal Pre-training for Document Understanding92.91LayoutMask (base)2023-05-30
GeoLayoutLM: Geometric Pre-training for Visual Information Extraction✓ Link92.86GeoLayoutLM2023-04-21
LayoutLMv3: Pre-training for Document AI with Unified Text and Image Masking✓ Link92.08LayoutLMv3 Large2022-04-18
Modeling Layout Reading Order as Ordering Relations for Visually-rich Document Understanding✓ Link91.84RORE (GeoLayoutLM)2024-09-29
StrucTexTv2: Masked Visual-Textual Prediction for Document Image Pre-training✓ Link91.82StrucTexTv2 (large)2023-03-01
XDoc: Unified Pre-training for Cross-Format Document Understanding✓ Link89.4XDoc1M2022-10-06
StrucTexTv2: Masked Visual-Textual Prediction for Document Image Pre-training✓ Link89.23StrucTexTv2 (small)2023-03-01
LiLT: A Simple yet Effective Language-Independent Layout Transformer for Structured Document Understanding✓ Link88.41LILT2022-02-28
Reading Order Matters: Information Extraction from Visually-rich Documents by Token Path Prediction✓ Link85.16TPP (LayoutMask)2023-10-17
LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding✓ Link84.2LayoutLMv2LARGE2020-12-29
DocTr: Document Transformer for Structured Information Extraction in Documents84DocTr2023-07-16
LayoutLMv2: Multi-modal Pre-training for Visually-Rich Document Understanding✓ Link82.76LayoutLMv2BASE2020-12-29
Doc2Graph: a Task Agnostic Document Understanding Framework based on Graph Neural Networks✓ Link82.25Doc2Graph2022-08-23
Check out how to contribute new results.