layoutxlm / README.md
baylitoo's picture
End of training
3fff79e
metadata
license: mit
tags:
  - generated_from_trainer
datasets:
  - funsd-layoutlmv3
model-index:
  - name: layoutxlm
    results: []

layoutxlm

This model is a fine-tuned version of SCUT-DLVCLab/lilt-roberta-en-base on the funsd-layoutlmv3 dataset. It achieves the following results on the evaluation set:

  • Loss: 1.5889
  • Answer: {'precision': 0.8761904761904762, 'recall': 0.9008567931456548, 'f1': 0.8883524441762222, 'number': 817}
  • Header: {'precision': 0.6666666666666666, 'recall': 0.5546218487394958, 'f1': 0.6055045871559633, 'number': 119}
  • Question: {'precision': 0.8883968113374667, 'recall': 0.9312906220984215, 'f1': 0.9093381686310064, 'number': 1077}
  • Overall Precision: 0.8728
  • Overall Recall: 0.8967
  • Overall F1: 0.8846
  • Overall Accuracy: 0.8115

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

Training results

Training Loss Epoch Step Validation Loss Answer Header Question Overall Precision Overall Recall Overall F1 Overall Accuracy
0.4322 10.53 200 0.9083 {'precision': 0.7704569606801275, 'recall': 0.8873929008567931, 'f1': 0.8248009101251422, 'number': 817} {'precision': 0.6162790697674418, 'recall': 0.44537815126050423, 'f1': 0.5170731707317073, 'number': 119} {'precision': 0.866852886405959, 'recall': 0.8644382544103992, 'f1': 0.8656438865643886, 'number': 1077} 0.8134 0.8490 0.8308 0.7863
0.0467 21.05 400 1.2942 {'precision': 0.8496583143507973, 'recall': 0.9130966952264382, 'f1': 0.88023598820059, 'number': 817} {'precision': 0.6585365853658537, 'recall': 0.453781512605042, 'f1': 0.5373134328358209, 'number': 119} {'precision': 0.8859964093357271, 'recall': 0.9164345403899722, 'f1': 0.9009584664536742, 'number': 1077} 0.8616 0.8877 0.8745 0.7966
0.015 31.58 600 1.2662 {'precision': 0.8574739281575898, 'recall': 0.9057527539779682, 'f1': 0.880952380952381, 'number': 817} {'precision': 0.5304347826086957, 'recall': 0.5126050420168067, 'f1': 0.5213675213675214, 'number': 119} {'precision': 0.8793879387938794, 'recall': 0.9071494893221913, 'f1': 0.8930530164533822, 'number': 1077} 0.8511 0.8833 0.8669 0.8114
0.0081 42.11 800 1.5223 {'precision': 0.8710462287104623, 'recall': 0.8763769889840881, 'f1': 0.8737034777303235, 'number': 817} {'precision': 0.5882352941176471, 'recall': 0.5882352941176471, 'f1': 0.5882352941176471, 'number': 119} {'precision': 0.8885844748858448, 'recall': 0.903435468895079, 'f1': 0.8959484346224678, 'number': 1077} 0.8639 0.8738 0.8689 0.8041
0.0033 52.63 1000 1.4361 {'precision': 0.8502304147465438, 'recall': 0.9033047735618115, 'f1': 0.8759643916913946, 'number': 817} {'precision': 0.6144578313253012, 'recall': 0.42857142857142855, 'f1': 0.504950495049505, 'number': 119} {'precision': 0.8767605633802817, 'recall': 0.924791086350975, 'f1': 0.9001355625847266, 'number': 1077} 0.8553 0.8867 0.8707 0.8156
0.0026 63.16 1200 1.4994 {'precision': 0.8615560640732265, 'recall': 0.9216646266829865, 'f1': 0.8905972797161442, 'number': 817} {'precision': 0.5981308411214953, 'recall': 0.5378151260504201, 'f1': 0.5663716814159291, 'number': 119} {'precision': 0.8945454545454545, 'recall': 0.9136490250696379, 'f1': 0.9039963252181902, 'number': 1077} 0.8654 0.8947 0.8798 0.8208
0.0016 73.68 1400 1.6091 {'precision': 0.858139534883721, 'recall': 0.9033047735618115, 'f1': 0.8801431127012522, 'number': 817} {'precision': 0.5980392156862745, 'recall': 0.5126050420168067, 'f1': 0.5520361990950226, 'number': 119} {'precision': 0.8947849954254345, 'recall': 0.9080779944289693, 'f1': 0.9013824884792625, 'number': 1077} 0.8647 0.8828 0.8736 0.8167
0.0009 84.21 1600 1.6010 {'precision': 0.859122401847575, 'recall': 0.9106487148102815, 'f1': 0.8841354723707664, 'number': 817} {'precision': 0.6741573033707865, 'recall': 0.5042016806722689, 'f1': 0.576923076923077, 'number': 119} {'precision': 0.8882931188561215, 'recall': 0.9229340761374187, 'f1': 0.9052823315118397, 'number': 1077} 0.8669 0.8932 0.8799 0.8049
0.0006 94.74 1800 1.5889 {'precision': 0.8761904761904762, 'recall': 0.9008567931456548, 'f1': 0.8883524441762222, 'number': 817} {'precision': 0.6666666666666666, 'recall': 0.5546218487394958, 'f1': 0.6055045871559633, 'number': 119} {'precision': 0.8883968113374667, 'recall': 0.9312906220984215, 'f1': 0.9093381686310064, 'number': 1077} 0.8728 0.8967 0.8846 0.8115
0.0004 105.26 2000 1.6126 {'precision': 0.8634772462077013, 'recall': 0.9057527539779682, 'f1': 0.8841099163679809, 'number': 817} {'precision': 0.6538461538461539, 'recall': 0.5714285714285714, 'f1': 0.6098654708520179, 'number': 119} {'precision': 0.894404332129964, 'recall': 0.9201485608170845, 'f1': 0.9070938215102976, 'number': 1077} 0.8695 0.8937 0.8814 0.8127
0.0004 115.79 2200 1.6606 {'precision': 0.8403648802736602, 'recall': 0.9020807833537332, 'f1': 0.8701298701298701, 'number': 817} {'precision': 0.6509433962264151, 'recall': 0.5798319327731093, 'f1': 0.6133333333333333, 'number': 119} {'precision': 0.8884826325411335, 'recall': 0.9025069637883009, 'f1': 0.8954398894518655, 'number': 1077} 0.8560 0.8833 0.8694 0.7906
0.0002 126.32 2400 1.6619 {'precision': 0.8378684807256236, 'recall': 0.9045287637698899, 'f1': 0.8699234844025897, 'number': 817} {'precision': 0.6836734693877551, 'recall': 0.5630252100840336, 'f1': 0.6175115207373272, 'number': 119} {'precision': 0.881981981981982, 'recall': 0.9090064995357474, 'f1': 0.8952903520804755, 'number': 1077} 0.8541 0.8867 0.8701 0.7929

Framework versions

  • Transformers 4.30.2
  • Pytorch 2.1.0.dev20230523+cu117
  • Datasets 2.13.0
  • Tokenizers 0.13.3