DonutInvoiceCzechV03R

This model is a fine-tuned version of TomasFAV/DonutInvoiceCzechV0R on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2443
• Accuracy: 0.9274
• F1: 0.9077

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: 9e-05
• train_batch_size: 4
• eval_batch_size: 1
• 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: 20
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
0.2874	1.0	46	0.1856	0.9007	0.8788
0.1328	2.0	92	0.2057	0.8800	0.8535
0.0790	3.0	138	0.1899	0.8992	0.8921
0.0493	4.0	184	0.2266	0.9103	0.8912
0.0391	5.0	230	0.2266	0.8962	0.8739
0.0271	6.0	276	0.2532	0.8840	0.8658
0.0238	7.0	322	0.2393	0.9016	0.8803
0.0211	8.0	368	0.2429	0.9090	0.8846
0.0210	9.0	414	0.2326	0.9266	0.8889
0.0184	10.0	460	0.2241	0.9216	0.9026
0.0109	11.0	506	0.2483	0.9075	0.8933
0.0037	12.0	552	0.2443	0.9274	0.9077
0.0023	13.0	598	0.2457	0.9269	0.8991
0.0057	14.0	644	0.2397	0.9278	0.9026
0.0024	15.0	690	0.2320	0.9346	0.9077
0.0008	16.0	736	0.2390	0.9344	0.9077
0.0015	17.0	782	0.2401	0.9350	0.9077
0.0042	18.0	828	0.2405	0.9346	0.9077
0.0016	19.0	874	0.2426	0.9322	0.9060
0.0035	20.0	920	0.2426	0.9322	0.9060

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.22.2