layoutlm-funsd1

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: 0.6653
• Answer: {'precision': 0.6705756929637526, 'recall': 0.7775030902348579, 'f1': 0.7200915855752718, 'number': 809}
• Header: {'precision': 0.30275229357798167, 'recall': 0.2773109243697479, 'f1': 0.28947368421052627, 'number': 119}
• Question: {'precision': 0.7173732335827099, 'recall': 0.8103286384976526, 'f1': 0.7610229276895942, 'number': 1065}
• Overall Precision: 0.6778
• Overall Recall: 0.7652
• Overall F1: 0.7188
• Overall Accuracy: 0.7992

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.8388	1.0	10	1.6345	{'precision': 0.010158013544018058, 'recall': 0.011124845488257108, 'f1': 0.010619469026548672, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.12983770287141075, 'recall': 0.09765258215962441, 'f1': 0.11146838156484459, 'number': 1065}	0.0670	0.0567	0.0614	0.3424
1.5101	2.0	20	1.3279	{'precision': 0.10227272727272728, 'recall': 0.08899876390605686, 'f1': 0.09517514871116987, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.3082191780821918, 'recall': 0.4225352112676056, 'f1': 0.3564356435643564, 'number': 1065}	0.2412	0.2619	0.2511	0.5546
1.196	3.0	30	1.0812	{'precision': 0.33375, 'recall': 0.3300370828182942, 'f1': 0.3318831572405221, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.4708233413269384, 'recall': 0.5530516431924882, 'f1': 0.5086355785837651, 'number': 1065}	0.4153	0.4295	0.4223	0.6283
0.957	4.0	40	0.8960	{'precision': 0.5760082730093071, 'recall': 0.688504326328801, 'f1': 0.6272522522522522, 'number': 809}	{'precision': 0.027777777777777776, 'recall': 0.008403361344537815, 'f1': 0.012903225806451613, 'number': 119}	{'precision': 0.6268939393939394, 'recall': 0.6215962441314554, 'f1': 0.6242338519566243, 'number': 1065}	0.5925	0.6121	0.6022	0.7315
0.7609	5.0	50	0.7756	{'precision': 0.608955223880597, 'recall': 0.7564894932014833, 'f1': 0.6747519294377067, 'number': 809}	{'precision': 0.11428571428571428, 'recall': 0.06722689075630252, 'f1': 0.08465608465608465, 'number': 119}	{'precision': 0.6362098138747885, 'recall': 0.7061032863849765, 'f1': 0.669336893635959, 'number': 1065}	0.6079	0.6884	0.6456	0.7649
0.634	6.0	60	0.7261	{'precision': 0.6207951070336392, 'recall': 0.7527812113720643, 'f1': 0.6804469273743018, 'number': 809}	{'precision': 0.24, 'recall': 0.15126050420168066, 'f1': 0.18556701030927833, 'number': 119}	{'precision': 0.6666666666666666, 'recall': 0.7380281690140845, 'f1': 0.7005347593582888, 'number': 1065}	0.6322	0.7090	0.6684	0.7783
0.5815	7.0	70	0.6992	{'precision': 0.6612377850162866, 'recall': 0.7527812113720643, 'f1': 0.7040462427745664, 'number': 809}	{'precision': 0.27586206896551724, 'recall': 0.20168067226890757, 'f1': 0.23300970873786409, 'number': 119}	{'precision': 0.6899841017488076, 'recall': 0.8150234741784037, 'f1': 0.7473095135600517, 'number': 1065}	0.6624	0.7531	0.7049	0.7906
0.5279	8.0	80	0.6827	{'precision': 0.6687631027253669, 'recall': 0.788627935723115, 'f1': 0.7237663074305162, 'number': 809}	{'precision': 0.3010752688172043, 'recall': 0.23529411764705882, 'f1': 0.2641509433962264, 'number': 119}	{'precision': 0.7285464098073555, 'recall': 0.7812206572769953, 'f1': 0.7539646579066607, 'number': 1065}	0.6843	0.7516	0.7164	0.7973
0.4907	9.0	90	0.6732	{'precision': 0.6609442060085837, 'recall': 0.761433868974042, 'f1': 0.707639287765652, 'number': 809}	{'precision': 0.3148148148148148, 'recall': 0.2857142857142857, 'f1': 0.29955947136563876, 'number': 119}	{'precision': 0.7145214521452146, 'recall': 0.8131455399061033, 'f1': 0.7606499780412823, 'number': 1065}	0.6732	0.7607	0.7143	0.7971
0.4734	10.0	100	0.6653	{'precision': 0.6705756929637526, 'recall': 0.7775030902348579, 'f1': 0.7200915855752718, 'number': 809}	{'precision': 0.30275229357798167, 'recall': 0.2773109243697479, 'f1': 0.28947368421052627, 'number': 119}	{'precision': 0.7173732335827099, 'recall': 0.8103286384976526, 'f1': 0.7610229276895942, 'number': 1065}	0.6778	0.7652	0.7188	0.7992

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.1+cu121
• Datasets 2.19.2
• Tokenizers 0.19.1