lilt-en-funsd / README.md
ssakha's picture
End of training
597aa6e verified
metadata
license: mit
base_model: SCUT-DLVCLab/lilt-roberta-en-base
tags:
  - generated_from_trainer
model-index:
  - name: lilt-en-funsd
    results: []

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: 1.4190
  • Answer: {'precision': 0.8649592549476135, 'recall': 0.9094247246022031, 'f1': 0.886634844868735, 'number': 817}
  • Header: {'precision': 0.6597938144329897, 'recall': 0.5378151260504201, 'f1': 0.5925925925925926, 'number': 119}
  • Question: {'precision': 0.8840970350404312, 'recall': 0.9136490250696379, 'f1': 0.8986301369863015, 'number': 1077}
  • Overall Precision: 0.8656
  • Overall Recall: 0.8897
  • Overall F1: 0.8775
  • Overall Accuracy: 0.8041

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
0.4199 10.53 200 1.0303 {'precision': 0.8458904109589042, 'recall': 0.9069767441860465, 'f1': 0.8753691671588896, 'number': 817} {'precision': 0.4470588235294118, 'recall': 0.6386554621848739, 'f1': 0.5259515570934256, 'number': 119} {'precision': 0.8797653958944281, 'recall': 0.8356545961002786, 'f1': 0.8571428571428572, 'number': 1077} 0.8299 0.8530 0.8413 0.7784
0.0485 21.05 400 1.3395 {'precision': 0.8127705627705628, 'recall': 0.9192166462668299, 'f1': 0.8627225732337737, 'number': 817} {'precision': 0.6463414634146342, 'recall': 0.44537815126050423, 'f1': 0.527363184079602, 'number': 119} {'precision': 0.8720292504570384, 'recall': 0.8857938718662952, 'f1': 0.878857669276831, 'number': 1077} 0.8371 0.8733 0.8549 0.7851
0.0154 31.58 600 1.2980 {'precision': 0.8578034682080925, 'recall': 0.9082007343941249, 'f1': 0.8822829964328182, 'number': 817} {'precision': 0.5742574257425742, 'recall': 0.48739495798319327, 'f1': 0.5272727272727273, 'number': 119} {'precision': 0.87322695035461, 'recall': 0.914577530176416, 'f1': 0.8934240362811792, 'number': 1077} 0.8524 0.8867 0.8692 0.8145
0.0076 42.11 800 1.3862 {'precision': 0.8296703296703297, 'recall': 0.9241126070991432, 'f1': 0.8743485813549509, 'number': 817} {'precision': 0.6206896551724138, 'recall': 0.453781512605042, 'f1': 0.5242718446601942, 'number': 119} {'precision': 0.8699472759226714, 'recall': 0.9192200557103064, 'f1': 0.8939051918735892, 'number': 1077} 0.8426 0.8937 0.8674 0.8016
0.0055 52.63 1000 1.4190 {'precision': 0.8649592549476135, 'recall': 0.9094247246022031, 'f1': 0.886634844868735, 'number': 817} {'precision': 0.6597938144329897, 'recall': 0.5378151260504201, 'f1': 0.5925925925925926, 'number': 119} {'precision': 0.8840970350404312, 'recall': 0.9136490250696379, 'f1': 0.8986301369863015, 'number': 1077} 0.8656 0.8897 0.8775 0.8041
0.0026 63.16 1200 1.5891 {'precision': 0.8293478260869566, 'recall': 0.9339045287637698, 'f1': 0.8785261945883708, 'number': 817} {'precision': 0.5922330097087378, 'recall': 0.5126050420168067, 'f1': 0.5495495495495496, 'number': 119} {'precision': 0.9082217973231358, 'recall': 0.8820798514391829, 'f1': 0.8949599623174752, 'number': 1077} 0.8574 0.8813 0.8692 0.8058
0.0027 73.68 1400 1.6258 {'precision': 0.8331466965285554, 'recall': 0.9106487148102815, 'f1': 0.8701754385964913, 'number': 817} {'precision': 0.5833333333333334, 'recall': 0.5294117647058824, 'f1': 0.5550660792951542, 'number': 119} {'precision': 0.8783542039355993, 'recall': 0.9117920148560817, 'f1': 0.894760820045558, 'number': 1077} 0.8443 0.8887 0.8659 0.7927
0.0009 84.21 1600 1.6324 {'precision': 0.8621495327102804, 'recall': 0.9033047735618115, 'f1': 0.8822474596533174, 'number': 817} {'precision': 0.6, 'recall': 0.5294117647058824, 'f1': 0.5625, 'number': 119} {'precision': 0.8733153638814016, 'recall': 0.9025069637883009, 'f1': 0.8876712328767123, 'number': 1077} 0.8549 0.8808 0.8676 0.7950
0.0007 94.74 1800 1.8058 {'precision': 0.8278145695364238, 'recall': 0.9179926560587516, 'f1': 0.8705745792222868, 'number': 817} {'precision': 0.5714285714285714, 'recall': 0.5042016806722689, 'f1': 0.5357142857142857, 'number': 119} {'precision': 0.9029495718363464, 'recall': 0.8811513463324049, 'f1': 0.8919172932330827, 'number': 1077} 0.8531 0.8738 0.8633 0.7927
0.0005 105.26 2000 1.8281 {'precision': 0.8543799772468714, 'recall': 0.9192166462668299, 'f1': 0.8856132075471698, 'number': 817} {'precision': 0.5238095238095238, 'recall': 0.5546218487394958, 'f1': 0.5387755102040817, 'number': 119} {'precision': 0.9050279329608939, 'recall': 0.9025069637883009, 'f1': 0.903765690376569, 'number': 1077} 0.8605 0.8887 0.8744 0.7915
0.0004 115.79 2200 1.6623 {'precision': 0.8643867924528302, 'recall': 0.8971848225214198, 'f1': 0.8804804804804804, 'number': 817} {'precision': 0.5426356589147286, 'recall': 0.5882352941176471, 'f1': 0.5645161290322581, 'number': 119} {'precision': 0.8898916967509025, 'recall': 0.9155060352831941, 'f1': 0.9025171624713959, 'number': 1077} 0.8580 0.8887 0.8731 0.8055
0.0003 126.32 2400 1.7066 {'precision': 0.8649592549476135, 'recall': 0.9094247246022031, 'f1': 0.886634844868735, 'number': 817} {'precision': 0.5689655172413793, 'recall': 0.5546218487394958, 'f1': 0.5617021276595745, 'number': 119} {'precision': 0.8988970588235294, 'recall': 0.9080779944289693, 'f1': 0.9034642032332563, 'number': 1077} 0.8662 0.8877 0.8768 0.8010

Framework versions

  • Transformers 4.39.3
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2