nifras365/OCR-Model-v3

layoutlm-receipts

This model is a fine-tuned version of microsoft/layoutlm-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0896
• Precision: 0.75
• Recall: 0.75
• F1: 0.75

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: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 25

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1
0.1758	1.0	8	0.1713	0.3529	0.6	0.4444
0.1438	2.0	16	0.2149	0.1111	0.15	0.1277
0.0494	3.0	24	0.2381	0.36	0.45	0.4000
0.042	4.0	32	0.1144	0.5455	0.6	0.5714
0.0236	5.0	40	0.0788	0.7	0.7	0.7
0.0111	6.0	48	0.0804	0.8333	0.75	0.7895
0.0114	7.0	56	0.0964	0.6667	0.7	0.6829
0.0031	8.0	64	0.0892	0.8333	0.75	0.7895
0.0065	9.0	72	0.1038	0.75	0.75	0.75
0.0014	10.0	80	0.1093	0.75	0.75	0.75
0.0045	11.0	88	0.0998	0.75	0.75	0.75
0.0027	12.0	96	0.0738	0.9444	0.85	0.8947
0.0008	13.0	104	0.0745	0.9444	0.85	0.8947
0.0029	14.0	112	0.1234	0.5833	0.7	0.6364
0.004	15.0	120	0.0865	0.6364	0.7	0.6667
0.0007	16.0	128	0.0888	0.8333	0.75	0.7895
0.0055	17.0	136	0.0934	0.75	0.75	0.75
0.0004	18.0	144	0.0854	0.8333	0.75	0.7895
0.0004	19.0	152	0.0846	0.8333	0.75	0.7895
0.0005	20.0	160	0.0843	0.8333	0.75	0.7895
0.0005	21.0	168	0.0852	0.8333	0.75	0.7895
0.0004	22.0	176	0.0862	0.8333	0.75	0.7895
0.0005	23.0	184	0.0875	0.8333	0.75	0.7895
0.0003	24.0	192	0.0892	0.75	0.75	0.75
0.0005	25.0	200	0.0896	0.75	0.75	0.75

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.0