DonutInvoiceCzechV3

This model is a fine-tuned version of naver-clova-ix/donut-base-finetuned-cord-v2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.2946
• Accuracy: 0.9152
• F1: 0.8838

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: 40
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1
3.9346	1.0	46	2.6485	0.0116	0.0064
1.3843	2.0	92	1.2891	0.2501	0.2407
0.8102	3.0	138	0.7001	0.5997	0.5336
0.2805	4.0	184	0.4553	0.6544	0.6571
0.1336	5.0	230	0.3413	0.8010	0.7771
0.1062	6.0	276	0.2924	0.7955	0.7831
0.0869	7.0	322	0.2980	0.8219	0.7988
0.0957	8.0	368	0.3558	0.8100	0.7938
0.0704	9.0	414	0.3160	0.8147	0.8055
0.0674	10.0	460	0.3314	0.8531	0.8247
0.0464	11.0	506	0.3728	0.8521	0.8146
0.0358	12.0	552	0.3211	0.8372	0.8079
0.0222	13.0	598	0.3009	0.8836	0.8420
0.0299	14.0	644	0.2888	0.8698	0.8362
0.0133	15.0	690	0.3496	0.8558	0.8459
0.0201	16.0	736	0.2847	0.8961	0.8665
0.0142	17.0	782	0.3228	0.9005	0.8652
0.0163	18.0	828	0.3359	0.8669	0.8310
0.0096	19.0	874	0.3167	0.8759	0.8488
0.0175	20.0	920	0.2905	0.8938	0.8687
0.0129	21.0	966	0.3119	0.8797	0.8570
0.0081	22.0	1012	0.3157	0.8780	0.8729
0.0036	23.0	1058	0.2950	0.9029	0.8731
0.0049	24.0	1104	0.3194	0.9048	0.8632
0.0034	25.0	1150	0.3091	0.8987	0.8650
0.0012	26.0	1196	0.2910	0.8968	0.8718
0.0049	27.0	1242	0.2924	0.9115	0.8769
0.0025	28.0	1288	0.2939	0.9040	0.8679
0.0014	29.0	1334	0.2946	0.9152	0.8838
0.0014	30.0	1380	0.3091	0.8989	0.8676
0.0004	31.0	1426	0.2930	0.8991	0.8637
0.0005	32.0	1472	0.2962	0.8977	0.8747
0.0008	33.0	1518	0.2922	0.8974	0.8665
0.0004	34.0	1564	0.2875	0.8982	0.8696
0.0004	35.0	1610	0.2895	0.8962	0.8665
0.0042	36.0	1656	0.2877	0.8944	0.8665
0.0004	37.0	1702	0.2879	0.8965	0.8701
0.0003	38.0	1748	0.2875	0.8984	0.8718
0.0003	39.0	1794	0.2879	0.8984	0.8718
0.0004	40.0	1840	0.2874	0.8984	0.8718

Framework versions

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