donut-base-vishnu

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

• Loss: 0.3250

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: 2e-05
• train_batch_size: 2
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
2.9387	1.0	127	2.0237
0.8381	2.0	254	1.3332
1.4923	3.0	381	1.1110
0.9061	4.0	508	0.9530
0.4627	5.0	635	0.9156
0.4305	6.0	762	0.7884
0.4383	7.0	889	0.6936
0.1852	8.0	1016	0.6715
0.2348	9.0	1143	0.6209
0.3975	10.0	1270	0.5614
0.1548	11.0	1397	0.5152
0.0377	12.0	1524	0.5135
0.043	13.0	1651	0.4759
0.0698	14.0	1778	0.4697
0.0292	15.0	1905	0.4243
0.0516	16.0	2032	0.4594
0.2062	17.0	2159	0.4332
0.0307	18.0	2286	0.4030
0.0775	19.0	2413	0.4069
0.0157	20.0	2540	0.4111
0.0137	21.0	2667	0.4072
0.0148	22.0	2794	0.3938
0.0454	23.0	2921	0.3789
0.0023	24.0	3048	0.3864
0.0033	25.0	3175	0.3750
0.0292	26.0	3302	0.3847
0.0087	27.0	3429	0.3592
0.0032	28.0	3556	0.3665
0.0048	29.0	3683	0.3372
0.0035	30.0	3810	0.3349
0.0197	31.0	3937	0.3591
0.0006	32.0	4064	0.3504
0.0016	33.0	4191	0.3450
0.0006	34.0	4318	0.3505
0.0046	35.0	4445	0.3332
0.0045	36.0	4572	0.3206
0.0006	37.0	4699	0.3361
0.0039	38.0	4826	0.3348
0.0059	39.0	4953	0.3328
0.0039	40.0	5080	0.3406
0.0014	41.0	5207	0.3250

Framework versions

• Transformers 4.35.0
• Pytorch 2.1.0+cu118
• Datasets 2.14.6
• Tokenizers 0.14.1