layoutlm-funsd

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: 0.7180
• Answer: {'precision': 0.596, 'recall': 0.7367119901112484, 'f1': 0.6589275843007186, 'number': 809}
• Header: {'precision': 0.08571428571428572, 'recall': 0.05042016806722689, 'f1': 0.06349206349206349, 'number': 119}
• Question: {'precision': 0.6859706362153344, 'recall': 0.7896713615023474, 'f1': 0.7341772151898733, 'number': 1065}
• Overall Precision: 0.6285
• Overall Recall: 0.7240
• Overall F1: 0.6729
• Overall Accuracy: 0.7704

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: 8
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 5
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
1.6774	1.0	19	1.3758	{'precision': 0.06839378238341969, 'recall': 0.0815822002472188, 'f1': 0.07440811724915444, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.332824427480916, 'recall': 0.40938967136150234, 'f1': 0.367157894736842, 'number': 1065}	0.2207	0.2519	0.2352	0.4928
1.169	2.0	38	0.9500	{'precision': 0.4467425025853154, 'recall': 0.5339925834363412, 'f1': 0.48648648648648646, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.557753164556962, 'recall': 0.6619718309859155, 'f1': 0.605410047230571, 'number': 1065}	0.5076	0.5705	0.5372	0.6799
0.8429	3.0	57	0.7922	{'precision': 0.5751953125, 'recall': 0.7280593325092707, 'f1': 0.6426623022367702, 'number': 809}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 119}	{'precision': 0.6494676494676495, 'recall': 0.7446009389671362, 'f1': 0.6937882764654418, 'number': 1065}	0.6051	0.6934	0.6462	0.7457
0.6915	4.0	76	0.7294	{'precision': 0.5885262116716122, 'recall': 0.7354758961681088, 'f1': 0.6538461538461539, 'number': 809}	{'precision': 0.05172413793103448, 'recall': 0.025210084033613446, 'f1': 0.03389830508474576, 'number': 119}	{'precision': 0.6642628205128205, 'recall': 0.7784037558685446, 'f1': 0.7168179853004755, 'number': 1065}	0.6159	0.7160	0.6622	0.7651
0.6221	5.0	95	0.7180	{'precision': 0.596, 'recall': 0.7367119901112484, 'f1': 0.6589275843007186, 'number': 809}	{'precision': 0.08571428571428572, 'recall': 0.05042016806722689, 'f1': 0.06349206349206349, 'number': 119}	{'precision': 0.6859706362153344, 'recall': 0.7896713615023474, 'f1': 0.7341772151898733, 'number': 1065}	0.6285	0.7240	0.6729	0.7704

Framework versions

• Transformers 4.43.4
• Pytorch 2.4.0+cpu
• Datasets 2.20.0
• Tokenizers 0.19.1