donut-classification-turbo

This model is a fine-tuned version of donut-base-finetuned-rvl-cdip on rvl-cdip-document-classification dataset. It achieves the following results on the evaluation set:

• Loss: 0.0499
• Accuracy: 93.34%

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 1e-05
• train_batch_size: 2
• eval_batch_size: 8
• 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: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
0.1319	0.5	2000	0.1200
0.1365	1.0	4000	0.0845
0.1203	1.5	6000	0.0751
0.1128	2.0	8000	0.0677
0.0734	2.5	10000	0.0541
0.0707	3.0	12000	0.0499

Framework versions

• Transformers 4.56.1
• Pytorch 2.8.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1