layoutlm-FUNSD-only

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on the funsd dataset. It achieves the following results on the evaluation set:

• Loss: 0.5883
• Eader: {'precision': 0.3877551020408163, 'recall': 0.2289156626506024, 'f1': 0.2878787878787879, 'number': 83}
• Nswer: {'precision': 0.4581673306772908, 'recall': 0.5609756097560976, 'f1': 0.5043859649122806, 'number': 205}
• Uestion: {'precision': 0.36981132075471695, 'recall': 0.42424242424242425, 'f1': 0.3951612903225806, 'number': 231}
• Overall Precision: 0.4106
• Overall Recall: 0.4470
• Overall F1: 0.4280
• Overall Accuracy: 0.7852

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 9
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Eader	Nswer	Uestion	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.3004	1.0	8	1.0817	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 83}	{'precision': 0.07280832095096583, 'recall': 0.23902439024390243, 'f1': 0.11161731207289294, 'number': 205}	{'precision': 0.06845238095238096, 'recall': 0.19913419913419914, 'f1': 0.1018826135105205, 'number': 231}	0.0706	0.1830	0.1019	0.6047
1.0289	2.0	16	0.8889	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 83}	{'precision': 0.1986754966887417, 'recall': 0.43902439024390244, 'f1': 0.2735562310030395, 'number': 205}	{'precision': 0.17155756207674944, 'recall': 0.329004329004329, 'f1': 0.22551928783382788, 'number': 231}	0.1853	0.3198	0.2346	0.6935
0.8399	3.0	24	0.7179	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 83}	{'precision': 0.2890855457227139, 'recall': 0.47804878048780486, 'f1': 0.36029411764705876, 'number': 205}	{'precision': 0.2375366568914956, 'recall': 0.35064935064935066, 'f1': 0.28321678321678323, 'number': 231}	0.2602	0.3449	0.2966	0.7429
0.7069	4.0	32	0.6412	{'precision': 0.13636363636363635, 'recall': 0.03614457831325301, 'f1': 0.05714285714285714, 'number': 83}	{'precision': 0.37318840579710144, 'recall': 0.5024390243902439, 'f1': 0.4282744282744283, 'number': 205}	{'precision': 0.3356164383561644, 'recall': 0.42424242424242425, 'f1': 0.37476099426386233, 'number': 231}	0.3458	0.3931	0.3679	0.7591
0.5901	5.0	40	0.6059	{'precision': 0.2564102564102564, 'recall': 0.12048192771084337, 'f1': 0.1639344262295082, 'number': 83}	{'precision': 0.3925925925925926, 'recall': 0.5170731707317073, 'f1': 0.4463157894736842, 'number': 205}	{'precision': 0.3726235741444867, 'recall': 0.42424242424242425, 'f1': 0.3967611336032389, 'number': 231}	0.3741	0.4123	0.3923	0.7735
0.5121	6.0	48	0.5797	{'precision': 0.3269230769230769, 'recall': 0.20481927710843373, 'f1': 0.2518518518518518, 'number': 83}	{'precision': 0.4351145038167939, 'recall': 0.5560975609756098, 'f1': 0.48822269807280516, 'number': 205}	{'precision': 0.3527272727272727, 'recall': 0.4199134199134199, 'f1': 0.383399209486166, 'number': 231}	0.3871	0.4393	0.4116	0.7865
0.4503	7.0	56	0.5941	{'precision': 0.36, 'recall': 0.21686746987951808, 'f1': 0.2706766917293233, 'number': 83}	{'precision': 0.4474708171206226, 'recall': 0.5609756097560976, 'f1': 0.4978354978354979, 'number': 205}	{'precision': 0.3619402985074627, 'recall': 0.4199134199134199, 'f1': 0.38877755511022044, 'number': 231}	0.4	0.4432	0.4205	0.7799
0.4114	8.0	64	0.5924	{'precision': 0.38, 'recall': 0.2289156626506024, 'f1': 0.28571428571428575, 'number': 83}	{'precision': 0.4453125, 'recall': 0.5560975609756098, 'f1': 0.4945770065075922, 'number': 205}	{'precision': 0.3656716417910448, 'recall': 0.42424242424242425, 'f1': 0.39278557114228463, 'number': 231}	0.4024	0.4451	0.4227	0.7827
0.3935	9.0	72	0.5883	{'precision': 0.3877551020408163, 'recall': 0.2289156626506024, 'f1': 0.2878787878787879, 'number': 83}	{'precision': 0.4581673306772908, 'recall': 0.5609756097560976, 'f1': 0.5043859649122806, 'number': 205}	{'precision': 0.36981132075471695, 'recall': 0.42424242424242425, 'f1': 0.3951612903225806, 'number': 231}	0.4106	0.4470	0.4280	0.7852

Framework versions

• Transformers 4.49.0
• Pytorch 2.6.0+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0