LayoutLMv3 fine-tuned on the FUNSD dataset. Code and results are available at the official GitHub repository of my Master Degree thesis .

Results obtained using seqeval in strict mode:

Precision Recall F1-score Variance (F1)
Answer 0.90 0.91 0.90 3e-5
Header 0.61 0.66 0.63 4e-4
Question 0.88 0.87 0.88 1e-4
Micro avg 0.87 0.88 0.87 3e-5
Macro avg 0.79 0.82 0.80 3e-5
Weighted avg 0.87 0.88 0.87 3e-5