donut-proto-sroie

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

• Loss: 2.9048

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: 20

Training results

Training Loss	Epoch	Step	Validation Loss
7.7533	1.0	73	7.3383
6.2747	2.0	146	5.8340
4.5766	3.0	219	4.4653
4.2866	4.0	292	3.6184
3.5846	5.0	365	3.3728
3.5065	6.0	438	3.2303
2.8412	7.0	511	3.1083
3.4091	8.0	584	3.0442
3.0191	9.0	657	3.0130
2.5351	10.0	730	2.9678
2.9271	11.0	803	2.9627
2.7935	12.0	876	2.9427
3.0205	13.0	949	2.9298
3.1783	14.0	1022	2.9204
3.0527	15.0	1095	2.9210
2.2231	16.0	1168	2.9251
2.1284	17.0	1241	2.9015
2.6874	18.0	1314	2.9099
2.5325	19.0	1387	2.9091
2.6132	20.0	1460	2.9048

Framework versions

• Transformers 4.34.1
• Pytorch 2.1.0+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1