layoutlm-FUNSDxSynthetic

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.6138
• Eader: {'precision': 0.4098360655737705, 'recall': 0.30120481927710846, 'f1': 0.34722222222222227, 'number': 83}
• Nswer: {'precision': 0.45525291828793774, 'recall': 0.5707317073170731, 'f1': 0.5064935064935064, 'number': 205}
• Uestion: {'precision': 0.3793103448275862, 'recall': 0.42857142857142855, 'f1': 0.4024390243902439, 'number': 231}
• Overall Precision: 0.4162
• Overall Recall: 0.4644
• Overall F1: 0.4390
• Overall Accuracy: 0.7750

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
1.198	1.0	12	1.0274	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 83}	{'precision': 0.11785714285714285, 'recall': 0.32195121951219513, 'f1': 0.17254901960784313, 'number': 205}	{'precision': 0.11732851985559567, 'recall': 0.2813852813852814, 'f1': 0.16560509554140126, 'number': 231}	0.1176	0.2524	0.1604	0.6381
0.9302	2.0	24	0.7826	{'precision': 0.16666666666666666, 'recall': 0.012048192771084338, 'f1': 0.02247191011235955, 'number': 83}	{'precision': 0.21844660194174756, 'recall': 0.43902439024390244, 'f1': 0.2917341977309562, 'number': 205}	{'precision': 0.2109375, 'recall': 0.35064935064935066, 'f1': 0.2634146341463415, 'number': 231}	0.2145	0.3314	0.2604	0.7183
0.7111	3.0	36	0.6407	{'precision': 0.1794871794871795, 'recall': 0.08433734939759036, 'f1': 0.11475409836065575, 'number': 83}	{'precision': 0.3432343234323432, 'recall': 0.5073170731707317, 'f1': 0.40944881889763785, 'number': 205}	{'precision': 0.303886925795053, 'recall': 0.3722943722943723, 'f1': 0.3346303501945525, 'number': 231}	0.3152	0.3796	0.3444	0.7782
0.5314	4.0	48	0.6422	{'precision': 0.21666666666666667, 'recall': 0.1566265060240964, 'f1': 0.18181818181818182, 'number': 83}	{'precision': 0.3985239852398524, 'recall': 0.526829268292683, 'f1': 0.45378151260504207, 'number': 205}	{'precision': 0.373015873015873, 'recall': 0.4069264069264069, 'f1': 0.38923395445134573, 'number': 231}	0.3688	0.4143	0.3902	0.7626
0.4782	5.0	60	0.5865	{'precision': 0.3114754098360656, 'recall': 0.2289156626506024, 'f1': 0.2638888888888889, 'number': 83}	{'precision': 0.4036363636363636, 'recall': 0.5414634146341464, 'f1': 0.4625000000000001, 'number': 205}	{'precision': 0.336996336996337, 'recall': 0.39826839826839827, 'f1': 0.3650793650793651, 'number': 231}	0.3645	0.4277	0.3936	0.7784
0.3789	6.0	72	0.6069	{'precision': 0.3220338983050847, 'recall': 0.2289156626506024, 'f1': 0.2676056338028169, 'number': 83}	{'precision': 0.4367816091954023, 'recall': 0.5560975609756098, 'f1': 0.4892703862660945, 'number': 205}	{'precision': 0.37401574803149606, 'recall': 0.41125541125541126, 'f1': 0.3917525773195876, 'number': 231}	0.3972	0.4393	0.4172	0.7696
0.3423	7.0	84	0.6048	{'precision': 0.375, 'recall': 0.25301204819277107, 'f1': 0.3021582733812949, 'number': 83}	{'precision': 0.42911877394636017, 'recall': 0.5463414634146342, 'f1': 0.48068669527896996, 'number': 205}	{'precision': 0.39215686274509803, 'recall': 0.4329004329004329, 'f1': 0.411522633744856, 'number': 231}	0.4073	0.4489	0.4271	0.7782
0.2995	8.0	96	0.6146	{'precision': 0.3709677419354839, 'recall': 0.27710843373493976, 'f1': 0.31724137931034485, 'number': 83}	{'precision': 0.43346007604562736, 'recall': 0.5560975609756098, 'f1': 0.48717948717948717, 'number': 205}	{'precision': 0.3787878787878788, 'recall': 0.4329004329004329, 'f1': 0.40404040404040403, 'number': 231}	0.4024	0.4566	0.4278	0.7758
0.2774	9.0	108	0.6138	{'precision': 0.4098360655737705, 'recall': 0.30120481927710846, 'f1': 0.34722222222222227, 'number': 83}	{'precision': 0.45525291828793774, 'recall': 0.5707317073170731, 'f1': 0.5064935064935064, 'number': 205}	{'precision': 0.3793103448275862, 'recall': 0.42857142857142855, 'f1': 0.4024390243902439, 'number': 231}	0.4162	0.4644	0.4390	0.7750

Framework versions

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