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: 1.5192
• Answer: {'precision': 0.8623024830699775, 'recall': 0.9351285189718482, 'f1': 0.897240164415737, 'number': 817}
• Header: {'precision': 0.5980392156862745, 'recall': 0.5126050420168067, 'f1': 0.5520361990950226, 'number': 119}
• Question: {'precision': 0.9070191431175935, 'recall': 0.9238625812441968, 'f1': 0.9153633854645814, 'number': 1077}
• Overall Precision: 0.8729
• Overall Recall: 0.9041
• Overall F1: 0.8882
• Overall Accuracy: 0.8317

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
0.41	10.5263	200	1.0931	{'precision': 0.8183856502242153, 'recall': 0.8935128518971848, 'f1': 0.8543007606787594, 'number': 817}	{'precision': 0.42138364779874216, 'recall': 0.5630252100840336, 'f1': 0.48201438848920863, 'number': 119}	{'precision': 0.8991185112634672, 'recall': 0.8523676880222841, 'f1': 0.8751191611058151, 'number': 1077}	0.8277	0.8520	0.8397	0.7869
0.0535	21.0526	400	1.2583	{'precision': 0.8495475113122172, 'recall': 0.9192166462668299, 'f1': 0.8830099941211051, 'number': 817}	{'precision': 0.5636363636363636, 'recall': 0.5210084033613446, 'f1': 0.5414847161572053, 'number': 119}	{'precision': 0.8898999090081893, 'recall': 0.9080779944289693, 'f1': 0.8988970588235294, 'number': 1077}	0.8557	0.8897	0.8724	0.8223
0.0132	31.5789	600	1.3993	{'precision': 0.8563348416289592, 'recall': 0.9265605875152999, 'f1': 0.8900646678424456, 'number': 817}	{'precision': 0.6116504854368932, 'recall': 0.5294117647058824, 'f1': 0.5675675675675675, 'number': 119}	{'precision': 0.9144486692015209, 'recall': 0.89322191272052, 'f1': 0.9037106622827619, 'number': 1077}	0.8740	0.8852	0.8796	0.8171
0.0078	42.1053	800	1.4683	{'precision': 0.8583042973286876, 'recall': 0.9045287637698899, 'f1': 0.8808104886769966, 'number': 817}	{'precision': 0.684931506849315, 'recall': 0.42016806722689076, 'f1': 0.5208333333333334, 'number': 119}	{'precision': 0.9023041474654377, 'recall': 0.9090064995357474, 'f1': 0.9056429232192413, 'number': 1077}	0.8757	0.8783	0.8770	0.8070
0.0035	52.6316	1000	1.4809	{'precision': 0.8633177570093458, 'recall': 0.9045287637698899, 'f1': 0.8834429169157203, 'number': 817}	{'precision': 0.6582278481012658, 'recall': 0.4369747899159664, 'f1': 0.5252525252525252, 'number': 119}	{'precision': 0.886443661971831, 'recall': 0.9350046425255338, 'f1': 0.9100768187980117, 'number': 1077}	0.8682	0.8932	0.8805	0.8184
0.0032	63.1579	1200	1.4947	{'precision': 0.8544018058690744, 'recall': 0.9265605875152999, 'f1': 0.889019377568996, 'number': 817}	{'precision': 0.5238095238095238, 'recall': 0.46218487394957986, 'f1': 0.4910714285714286, 'number': 119}	{'precision': 0.9100185528756958, 'recall': 0.9108635097493036, 'f1': 0.9104408352668213, 'number': 1077}	0.8666	0.8907	0.8785	0.8247
0.0016	73.6842	1400	1.4909	{'precision': 0.8579676674364896, 'recall': 0.9094247246022031, 'f1': 0.8829471182412357, 'number': 817}	{'precision': 0.6666666666666666, 'recall': 0.5378151260504201, 'f1': 0.5953488372093023, 'number': 119}	{'precision': 0.9136822773186409, 'recall': 0.9238625812441968, 'f1': 0.9187442289935365, 'number': 1077}	0.8786	0.8952	0.8868	0.8234
0.0006	84.2105	1600	1.5053	{'precision': 0.8689492325855962, 'recall': 0.9008567931456548, 'f1': 0.8846153846153847, 'number': 817}	{'precision': 0.5922330097087378, 'recall': 0.5126050420168067, 'f1': 0.5495495495495496, 'number': 119}	{'precision': 0.8995475113122172, 'recall': 0.9229340761374187, 'f1': 0.9110907424381303, 'number': 1077}	0.8715	0.8897	0.8805	0.8269
0.0005	94.7368	1800	1.5094	{'precision': 0.8648648648648649, 'recall': 0.9400244798041616, 'f1': 0.9008797653958945, 'number': 817}	{'precision': 0.6138613861386139, 'recall': 0.5210084033613446, 'f1': 0.5636363636363637, 'number': 119}	{'precision': 0.9150141643059491, 'recall': 0.8997214484679665, 'f1': 0.9073033707865169, 'number': 1077}	0.8784	0.8937	0.8860	0.8309
0.0004	105.2632	2000	1.5111	{'precision': 0.8807017543859649, 'recall': 0.9216646266829865, 'f1': 0.9007177033492823, 'number': 817}	{'precision': 0.61, 'recall': 0.5126050420168067, 'f1': 0.5570776255707762, 'number': 119}	{'precision': 0.8981064021641119, 'recall': 0.924791086350975, 'f1': 0.9112534309240622, 'number': 1077}	0.8769	0.8992	0.8879	0.8322
0.0004	115.7895	2200	1.5100	{'precision': 0.8672768878718535, 'recall': 0.9277845777233782, 'f1': 0.8965109402720284, 'number': 817}	{'precision': 0.6145833333333334, 'recall': 0.4957983193277311, 'f1': 0.5488372093023256, 'number': 119}	{'precision': 0.9016245487364621, 'recall': 0.9275766016713092, 'f1': 0.91441647597254, 'number': 1077}	0.8739	0.9021	0.8878	0.8312
0.0002	126.3158	2400	1.5192	{'precision': 0.8623024830699775, 'recall': 0.9351285189718482, 'f1': 0.897240164415737, 'number': 817}	{'precision': 0.5980392156862745, 'recall': 0.5126050420168067, 'f1': 0.5520361990950226, 'number': 119}	{'precision': 0.9070191431175935, 'recall': 0.9238625812441968, 'f1': 0.9153633854645814, 'number': 1077}	0.8729	0.9041	0.8882	0.8317

Framework versions

• Transformers 4.43.3
• Pytorch 2.0.0+cu118
• Datasets 2.20.0
• Tokenizers 0.19.1