Benchmark Stats
SOTA History
f1
f1
Higher is better
| Rank | Model | Source | Score | Year | Paper |
|---|---|---|---|---|---|
| 1 | LayoutLMv3-large LayoutLMv3-large. Table 1 in paper. ACM MM 2022. SOTA at time of publication. | Community | 92.08 | 2022 | Source |
| 2 | UDOP UDOP (Unified Document Processing). Table 3 in paper. CVPR 2023. Single generative model for all document tasks. | Community | 91.62 | 2023 | Source |
| 3 | LayoutLMv3-base LayoutLMv3-base. Table 1 in paper. ACM MM 2022. | Community | 90.29 | 2022 | Source |
| 4 | DocFormerv2-large DocFormerv2-large. Table 5 in paper. ICCV 2023. | Community | 88.89 | 2023 | Source |
| 5 | LiLT[EN-R2]-base LiLT with English RoBERTa backbone (EN-R2), base size. Table 2 in paper. ACL 2022. Best monolingual FUNSD result. | Community | 88.41 | 2022 | Source |
| 6 | DocFormerv2-base DocFormerv2-base. Table 5 in paper. ICCV 2023. | Community | 88.37 | 2023 | Source |
| 7 | StructuralLM StructuralLM (large). Table 1 in paper. ACL 2021. Precision 83.52, Recall 86.81. | Community | 85.14 | 2021 | Source |
| 8 | FormNet FormNet. Table 1 in paper. ACL 2022. Uses rich structural encoding via graph neural network. | Community | 84.69 | 2022 | Source |
| 9 | BROS-large BROS-large on FUNSD entity extraction. Table 3 in paper. AAAI 2022. | Community | 84.52 | 2022 | Source |
| 10 | LayoutLMv2-large LayoutLMv2-large. Table 6 in paper. ACL 2021. | Community | 84.2 | 2021 | Source |
| 11 | LayoutLMv2-base LayoutLMv2-base. Table 6 in paper. ACL 2021. | Community | 82.76 | 2021 | Source |
| 12 | LayoutLMv1-base LayoutLM-base with text+layout+image embeddings, 11M docs. Best base variant. Table 1 in paper. ACL 2020. | Community | 79.27 | 2020 | Source |
| 13 | LayoutLMv1-large LayoutLM-large, text+layout, MVLM, 11M docs 1 epoch. Table 1 in paper. ACL 2020. | Community | 77.89 | 2020 | Source |