donut_experiment_bayesian_trial_16

This model is a fine-tuned version of naver-clova-ix/donut-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5541
• Bleu: 0.0670
• Precisions: [0.8417721518987342, 0.7841726618705036, 0.7388888888888889, 0.6996699669966997]
• Brevity Penalty: 0.0876
• Length Ratio: 0.2912
• Translation Length: 474
• Reference Length: 1628
• Cer: 0.7567
• Wer: 0.8224

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: 0.00011219603369833024
• train_batch_size: 1
• eval_batch_size: 1
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 2
• 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
0.0965	1.0	253	0.5550	0.0624	[0.7995824634655533, 0.7085308056872038, 0.6520547945205479, 0.6038961038961039]	0.0908	0.2942	479	1628	0.7576	0.8347
0.0844	2.0	506	0.5896	0.0651	[0.8218029350104822, 0.7476190476190476, 0.696969696969697, 0.6535947712418301]	0.0895	0.2930	477	1628	0.7557	0.8302
0.0539	3.0	759	0.5594	0.0666	[0.8322851153039832, 0.7642857142857142, 0.7134986225895317, 0.673202614379085]	0.0895	0.2930	477	1628	0.7552	0.8223
0.023	4.0	1012	0.5541	0.0670	[0.8417721518987342, 0.7841726618705036, 0.7388888888888889, 0.6996699669966997]	0.0876	0.2912	474	1628	0.7567	0.8224

Framework versions

• Transformers 4.40.0
• Pytorch 2.1.0
• Datasets 2.18.0
• Tokenizers 0.19.1