donut-base-sroie-v3

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: 1.7091
• Bleu: 0.0045
• Precisions: [0.32051282051282054, 0.10617283950617284, 0.043859649122807015, 0.02867383512544803]
• Brevity Penalty: 0.0555
• Length Ratio: 0.2570
• Translation Length: 468
• Reference Length: 1821
• Cer: 0.8657
• Wer: 0.9978

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: 4
• eval_batch_size: 4
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Precisions	Brevity Penalty	Length Ratio	Translation Length	Reference Length	Cer	Wer
No log	0.99	62	1.7048	0.0026	[0.3287981859410431, 0.09259259259259259, 0.025396825396825397, 0.015873015873015872]	0.0438	0.2422	441	1821	0.8729	1.0
0.6536	2.0	125	1.7425	0.0035	[0.32051282051282054, 0.0962962962962963, 0.029239766081871343, 0.017921146953405017]	0.0555	0.2570	468	1821	0.8701	0.9986
0.6536	2.99	187	1.6949	0.0038	[0.3148936170212766, 0.09582309582309582, 0.03197674418604651, 0.021352313167259787]	0.0564	0.2581	470	1821	0.8670	0.9978
0.6585	3.97	248	1.7091	0.0045	[0.32051282051282054, 0.10617283950617284, 0.043859649122807015, 0.02867383512544803]	0.0555	0.2570	468	1821	0.8657	0.9978

Framework versions

• Transformers 4.40.0.dev0
• Pytorch 2.1.0
• Datasets 2.18.0
• Tokenizers 0.15.2