DonutInvoiceCzechV013R

This model is a fine-tuned version of TomasFAV/DonutInvoiceCzechV01R on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2355
• Accuracy: 0.9150
• F1: 0.9043

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
0.2561	1.0	46	0.1958	0.8819	0.8581
0.1479	2.0	92	0.1839	0.9026	0.8821
0.0730	3.0	138	0.1941	0.9106	0.8889
0.0482	4.0	184	0.2087	0.9053	0.8872
0.0209	5.0	230	0.2163	0.8984	0.8691
0.0270	6.0	276	0.2235	0.9172	0.8821
0.0131	7.0	322	0.2250	0.9142	0.8862
0.0361	8.0	368	0.2241	0.9129	0.8849
0.0087	9.0	414	0.2465	0.9170	0.8906
0.0188	10.0	460	0.2351	0.9108	0.8855
0.0092	11.0	506	0.2228	0.9138	0.8889
0.0059	12.0	552	0.2324	0.9175	0.8974
0.0010	13.0	598	0.2355	0.9150	0.9043
0.0024	14.0	644	0.2377	0.9208	0.8991
0.0013	15.0	690	0.2408	0.9183	0.8974
0.0007	16.0	736	0.2449	0.9094	0.8906
0.0019	17.0	782	0.2480	0.9159	0.8991
0.0004	18.0	828	0.2466	0.9110	0.8906
0.0004	19.0	874	0.2475	0.9126	0.8923
0.0006	20.0	920	0.2474	0.9130	0.8923

Framework versions

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