lilt-en-funsd

This model is a fine-tuned version of SCUT-DLVCLab/lilt-roberta-en-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.6973
• Answer: {'precision': 0.8658109684947491, 'recall': 0.9082007343941249, 'f1': 0.886499402628435, 'number': 817}
• Header: {'precision': 0.6770833333333334, 'recall': 0.5462184873949579, 'f1': 0.6046511627906976, 'number': 119}
• Question: {'precision': 0.9074243813015582, 'recall': 0.9192200557103064, 'f1': 0.9132841328413284, 'number': 1077}
• Overall Precision: 0.8792
• Overall Recall: 0.8927
• Overall F1: 0.8859
• Overall Accuracy: 0.8011

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 2500
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Answer	Header	Question	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.1857	26.32	500	1.4181	{'precision': 0.8298109010011123, 'recall': 0.9130966952264382, 'f1': 0.8694638694638694, 'number': 817}	{'precision': 0.6923076923076923, 'recall': 0.5294117647058824, 'f1': 0.5999999999999999, 'number': 119}	{'precision': 0.886672710788758, 'recall': 0.9080779944289693, 'f1': 0.8972477064220182, 'number': 1077}	0.8538	0.8877	0.8704	0.7981
0.0068	52.63	1000	1.6084	{'precision': 0.8581235697940504, 'recall': 0.9179926560587516, 'f1': 0.8870490833826139, 'number': 817}	{'precision': 0.5877192982456141, 'recall': 0.5630252100840336, 'f1': 0.5751072961373391, 'number': 119}	{'precision': 0.9083255378858747, 'recall': 0.9015784586815228, 'f1': 0.9049394221808015, 'number': 1077}	0.8692	0.8882	0.8786	0.7956
0.0018	78.95	1500	1.6068	{'precision': 0.8742655699177438, 'recall': 0.9106487148102815, 'f1': 0.8920863309352519, 'number': 817}	{'precision': 0.6050420168067226, 'recall': 0.6050420168067226, 'f1': 0.6050420168067226, 'number': 119}	{'precision': 0.902867715078631, 'recall': 0.9062209842154132, 'f1': 0.9045412418906396, 'number': 1077}	0.8737	0.8902	0.8819	0.8095
0.0007	105.26	2000	1.6522	{'precision': 0.8611111111111112, 'recall': 0.9106487148102815, 'f1': 0.8851873884592504, 'number': 817}	{'precision': 0.6126126126126126, 'recall': 0.5714285714285714, 'f1': 0.591304347826087, 'number': 119}	{'precision': 0.9098513011152416, 'recall': 0.9090064995357474, 'f1': 0.9094287041337669, 'number': 1077}	0.8732	0.8897	0.8814	0.8028
0.0002	131.58	2500	1.6973	{'precision': 0.8658109684947491, 'recall': 0.9082007343941249, 'f1': 0.886499402628435, 'number': 817}	{'precision': 0.6770833333333334, 'recall': 0.5462184873949579, 'f1': 0.6046511627906976, 'number': 119}	{'precision': 0.9074243813015582, 'recall': 0.9192200557103064, 'f1': 0.9132841328413284, 'number': 1077}	0.8792	0.8927	0.8859	0.8011

Framework versions

• Transformers 4.36.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.0
• Tokenizers 0.15.0