metadata

license: mit
tags:
  - generated_from_trainer
datasets:
  - funsd-layoutlmv3
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 the funsd-layoutlmv3 dataset. It achieves the following results on the evaluation set:

Loss: 1.5849
Answer: {'precision': 0.8488636363636364, 'recall': 0.9143206854345165, 'f1': 0.8803771361225692, 'number': 817}
Header: {'precision': 0.5631067961165048, 'recall': 0.48739495798319327, 'f1': 0.5225225225225225, 'number': 119}
Question: {'precision': 0.8887859128822985, 'recall': 0.8904363974001857, 'f1': 0.8896103896103896, 'number': 1077}
Overall Precision: 0.8555
Overall Recall: 0.8763
Overall F1: 0.8658
Overall Accuracy: 0.8017

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: 2
eval_batch_size: 2
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
0.7834	2.67	200	0.5897	{'precision': 0.8059536934950385, 'recall': 0.8947368421052632, 'f1': 0.8480278422273781, 'number': 817}	{'precision': 0.4069767441860465, 'recall': 0.29411764705882354, 'f1': 0.34146341463414637, 'number': 119}	{'precision': 0.8141447368421053, 'recall': 0.9192200557103064, 'f1': 0.8634976013955518, 'number': 1077}	0.7949	0.8723	0.8318	0.7937
0.2848	5.33	400	0.8788	{'precision': 0.8155136268343816, 'recall': 0.9522643818849449, 'f1': 0.8785996612083569, 'number': 817}	{'precision': 0.46218487394957986, 'recall': 0.46218487394957986, 'f1': 0.46218487394957986, 'number': 119}	{'precision': 0.8724584103512015, 'recall': 0.8765088207985144, 'f1': 0.8744789254284391, 'number': 1077}	0.8246	0.8828	0.8527	0.7916
0.1446	8.0	600	1.0402	{'precision': 0.8363844393592678, 'recall': 0.8947368421052632, 'f1': 0.8645771732702543, 'number': 817}	{'precision': 0.5894736842105263, 'recall': 0.47058823529411764, 'f1': 0.5233644859813084, 'number': 119}	{'precision': 0.8886861313868614, 'recall': 0.904363974001857, 'f1': 0.8964565117349288, 'number': 1077}	0.8528	0.8748	0.8637	0.7891
0.0791	10.67	800	1.1878	{'precision': 0.8659549228944247, 'recall': 0.8935128518971848, 'f1': 0.8795180722891566, 'number': 817}	{'precision': 0.48360655737704916, 'recall': 0.4957983193277311, 'f1': 0.4896265560165975, 'number': 119}	{'precision': 0.8647007805724197, 'recall': 0.9257195914577531, 'f1': 0.8941704035874439, 'number': 1077}	0.8432	0.8872	0.8647	0.7973
0.0459	13.33	1000	1.3884	{'precision': 0.8078175895765473, 'recall': 0.9106487148102815, 'f1': 0.856156501726122, 'number': 817}	{'precision': 0.5789473684210527, 'recall': 0.46218487394957986, 'f1': 0.514018691588785, 'number': 119}	{'precision': 0.8819702602230484, 'recall': 0.8811513463324049, 'f1': 0.8815606130980028, 'number': 1077}	0.8356	0.8684	0.8516	0.7831
0.0278	16.0	1200	1.4707	{'precision': 0.8415051311288484, 'recall': 0.9033047735618115, 'f1': 0.8713105076741442, 'number': 817}	{'precision': 0.5869565217391305, 'recall': 0.453781512605042, 'f1': 0.5118483412322274, 'number': 119}	{'precision': 0.8690685413005272, 'recall': 0.9182915506035283, 'f1': 0.8930022573363431, 'number': 1077}	0.8453	0.8847	0.8646	0.7801
0.0105	18.67	1400	1.5749	{'precision': 0.843069873997709, 'recall': 0.9008567931456548, 'f1': 0.8710059171597634, 'number': 817}	{'precision': 0.5426356589147286, 'recall': 0.5882352941176471, 'f1': 0.5645161290322581, 'number': 119}	{'precision': 0.8799630655586335, 'recall': 0.8848653667595172, 'f1': 0.8824074074074075, 'number': 1077}	0.8436	0.8738	0.8585	0.7800
0.0083	21.33	1600	1.5530	{'precision': 0.8551236749116607, 'recall': 0.8886168910648715, 'f1': 0.8715486194477792, 'number': 817}	{'precision': 0.5727272727272728, 'recall': 0.5294117647058824, 'f1': 0.5502183406113538, 'number': 119}	{'precision': 0.8669032830523514, 'recall': 0.9071494893221913, 'f1': 0.8865698729582577, 'number': 1077}	0.8466	0.8773	0.8617	0.8007
0.0045	24.0	1800	1.5849	{'precision': 0.8488636363636364, 'recall': 0.9143206854345165, 'f1': 0.8803771361225692, 'number': 817}	{'precision': 0.5631067961165048, 'recall': 0.48739495798319327, 'f1': 0.5225225225225225, 'number': 119}	{'precision': 0.8887859128822985, 'recall': 0.8904363974001857, 'f1': 0.8896103896103896, 'number': 1077}	0.8555	0.8763	0.8658	0.8017
0.0025	26.67	2000	1.6119	{'precision': 0.8464203233256351, 'recall': 0.8971848225214198, 'f1': 0.8710635769459298, 'number': 817}	{'precision': 0.5566037735849056, 'recall': 0.4957983193277311, 'f1': 0.5244444444444444, 'number': 119}	{'precision': 0.8709386281588448, 'recall': 0.8960074280408542, 'f1': 0.8832951945080092, 'number': 1077}	0.8447	0.8728	0.8585	0.7914
0.0019	29.33	2200	1.5579	{'precision': 0.8661971830985915, 'recall': 0.9033047735618115, 'f1': 0.884361893349311, 'number': 817}	{'precision': 0.5660377358490566, 'recall': 0.5042016806722689, 'f1': 0.5333333333333334, 'number': 119}	{'precision': 0.8654188948306596, 'recall': 0.9015784586815228, 'f1': 0.8831286948613006, 'number': 1077}	0.8505	0.8788	0.8644	0.7971
0.0008	32.0	2400	1.5983	{'precision': 0.8530424799081515, 'recall': 0.9094247246022031, 'f1': 0.8803317535545023, 'number': 817}	{'precision': 0.5714285714285714, 'recall': 0.5042016806722689, 'f1': 0.5357142857142857, 'number': 119}	{'precision': 0.8768248175182481, 'recall': 0.8922934076137419, 'f1': 0.8844914864242981, 'number': 1077}	0.8514	0.8763	0.8636	0.7982

Framework versions

Transformers 4.30.0.dev0
Pytorch 1.8.0+cu101
Datasets 2.12.0
Tokenizers 0.13.3