layoutlmv3-base-cord

This model is a fine-tuned version of microsoft/layoutlmv3-base on the mp-02/cord dataset. It achieves the following results on the evaluation set:

• Loss: 0.1517
• Precision: 0.9752
• Recall: 0.9785
• F1: 0.9768
• Accuracy: 0.9739

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 3000

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	2.0	100	0.8667	0.7592	0.8202	0.7885	0.8097
No log	4.0	200	0.3443	0.9122	0.9387	0.9253	0.9222
No log	6.0	300	0.2128	0.9345	0.9569	0.9456	0.9579
No log	8.0	400	0.1745	0.9440	0.9635	0.9537	0.9629
0.6362	10.0	500	0.1594	0.9559	0.9702	0.9630	0.9684
0.6362	12.0	600	0.1720	0.9630	0.9693	0.9661	0.9629
0.6362	14.0	700	0.1528	0.9607	0.9710	0.9658	0.9675
0.6362	16.0	800	0.1460	0.9638	0.9718	0.9678	0.9680
0.6362	18.0	900	0.1609	0.9614	0.9702	0.9658	0.9648
0.0536	20.0	1000	0.1517	0.9752	0.9785	0.9768	0.9739
0.0536	22.0	1100	0.1901	0.9614	0.9693	0.9653	0.9657
0.0536	24.0	1200	0.1867	0.9638	0.9718	0.9678	0.9666

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.0+cu118
• Datasets 2.21.0
• Tokenizers 0.19.1