layoutlm-doclaynet-test

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

• Loss: 0.3029
• Footer: {'precision': 0.7619047619047619, 'recall': 0.7960199004975125, 'f1': 0.7785888077858881, 'number': 201}
• Header: {'precision': 0.7631578947368421, 'recall': 0.6987951807228916, 'f1': 0.7295597484276729, 'number': 83}
• Able: {'precision': 0.569377990430622, 'recall': 0.7531645569620253, 'f1': 0.6485013623978202, 'number': 158}
• Aption: {'precision': 0.2857142857142857, 'recall': 0.26865671641791045, 'f1': 0.2769230769230769, 'number': 67}
• Ext: {'precision': 0.6098901098901099, 'recall': 0.6809815950920245, 'f1': 0.6434782608695652, 'number': 326}
• Icture: {'precision': 0.18055555555555555, 'recall': 0.2, 'f1': 0.18978102189781024, 'number': 65}
• Itle: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}
• Ootnote: {'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 4}
• Overall Precision: 0.5930
• Overall Recall: 0.6505
• Overall F1: 0.6204
• Overall Accuracy: 0.9197

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: 3

Training results

Training Loss	Epoch	Step	Validation Loss	Footer	Header	Able	Aption	Ext	Icture	Itle	Ootnote	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.2414	1.0	426	0.1727	{'precision': 0.6724137931034483, 'recall': 0.7761194029850746, 'f1': 0.720554272517321, 'number': 201}	{'precision': 0.7142857142857143, 'recall': 0.5421686746987951, 'f1': 0.6164383561643836, 'number': 83}	{'precision': 0.5069124423963134, 'recall': 0.6962025316455697, 'f1': 0.5866666666666668, 'number': 158}	{'precision': 0.22916666666666666, 'recall': 0.16417910447761194, 'f1': 0.19130434782608696, 'number': 67}	{'precision': 0.5323383084577115, 'recall': 0.656441717791411, 'f1': 0.587912087912088, 'number': 326}	{'precision': 0.24528301886792453, 'recall': 0.2, 'f1': 0.22033898305084745, 'number': 65}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 4}	0.5409	0.6053	0.5713	0.9584
0.1037	2.0	852	0.1726	{'precision': 0.7045454545454546, 'recall': 0.7711442786069652, 'f1': 0.7363420427553445, 'number': 201}	{'precision': 0.8529411764705882, 'recall': 0.6987951807228916, 'f1': 0.7682119205298014, 'number': 83}	{'precision': 0.5658536585365853, 'recall': 0.7341772151898734, 'f1': 0.6391184573002755, 'number': 158}	{'precision': 0.25333333333333335, 'recall': 0.2835820895522388, 'f1': 0.2676056338028169, 'number': 67}	{'precision': 0.5640394088669951, 'recall': 0.7024539877300614, 'f1': 0.6256830601092896, 'number': 326}	{'precision': 0.16666666666666666, 'recall': 0.18461538461538463, 'f1': 0.17518248175182485, 'number': 65}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 4}	0.5631	0.6494	0.6032	0.9510
0.0647	3.0	1278	0.3029	{'precision': 0.7619047619047619, 'recall': 0.7960199004975125, 'f1': 0.7785888077858881, 'number': 201}	{'precision': 0.7631578947368421, 'recall': 0.6987951807228916, 'f1': 0.7295597484276729, 'number': 83}	{'precision': 0.569377990430622, 'recall': 0.7531645569620253, 'f1': 0.6485013623978202, 'number': 158}	{'precision': 0.2857142857142857, 'recall': 0.26865671641791045, 'f1': 0.2769230769230769, 'number': 67}	{'precision': 0.6098901098901099, 'recall': 0.6809815950920245, 'f1': 0.6434782608695652, 'number': 326}	{'precision': 0.18055555555555555, 'recall': 0.2, 'f1': 0.18978102189781024, 'number': 65}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 3}	{'precision': 0.0, 'recall': 0.0, 'f1': 0.0, 'number': 4}	0.5930	0.6505	0.6204	0.9197

Framework versions

• Transformers 4.26.1
• Pytorch 1.12.1+cu102
• Datasets 2.9.0
• Tokenizers 0.13.2