layoutlm-funsd-sequence-tf

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on an unknown dataset. It achieves the following results on the evaluation set:

• Train Loss: 0.2348
• Validation Loss: 0.6737
• Train Overall Precision: 0.7356
• Train Overall Recall: 0.7998
• Train Overall F1: 0.7663
• Train Overall Accuracy: 0.8220
• Epoch: 7

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:

• optimizer: {'inner_optimizer': {'class_name': 'AdamWeightDecay', 'config': {'name': 'AdamWeightDecay', 'learning_rate': 3e-05, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-07, 'amsgrad': False, 'weight_decay_rate': 0.01}}, 'dynamic': True, 'initial_scale': 32768.0, 'dynamic_growth_steps': 2000}
• training_precision: mixed_float16

Training results

Train Loss	Validation Loss	Train Overall Precision	Train Overall Recall	Train Overall F1	Train Overall Accuracy	Epoch
1.7150	1.4139	0.2373	0.2860	0.2594	0.4954	0
1.1803	0.9205	0.5676	0.6322	0.5981	0.7008	1
0.7884	0.7100	0.6202	0.7250	0.6685	0.7735	2
0.5877	0.6476	0.6689	0.7662	0.7142	0.7942	3
0.4490	0.6179	0.7133	0.8078	0.7576	0.8066	4
0.3746	0.6305	0.7176	0.7878	0.7510	0.8129	5
0.3082	0.6924	0.7163	0.8018	0.7566	0.7937	6
0.2348	0.6737	0.7356	0.7998	0.7663	0.8220	7

Framework versions

• Transformers 4.28.1
• TensorFlow 2.12.0
• Datasets 2.11.0
• Tokenizers 0.13.3