layoutlmv3-funsd

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

• Loss: 0.8428
• Precision: 0.8993
• Recall: 0.9046
• F1: 0.9019
• Accuracy: 0.8354

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: 1e-05
• train_batch_size: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 1000
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	2.63	100	0.6294	0.7864	0.8286	0.8070	0.7966
No log	5.26	200	0.5034	0.8389	0.8793	0.8586	0.8343
No log	7.89	300	0.5673	0.8597	0.9011	0.8799	0.8416
No log	10.53	400	0.5730	0.8783	0.9106	0.8941	0.8395
0.4463	13.16	500	0.6630	0.8923	0.9016	0.8970	0.8412
0.4463	15.79	600	0.7048	0.8850	0.8947	0.8898	0.8329
0.4463	18.42	700	0.7772	0.8925	0.9071	0.8997	0.8317
0.4463	21.05	800	0.8408	0.8959	0.9016	0.8987	0.8313
0.4463	23.68	900	0.8580	0.8918	0.9051	0.8984	0.8313
0.0611	26.32	1000	0.8428	0.8993	0.9046	0.9019	0.8354

Framework versions

• Transformers 4.36.0
• Pytorch 2.0.0
• Datasets 2.16.1
• Tokenizers 0.15.0