donut_experiment_bayesian_trial_2

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: 0.4983
• Bleu: 0.0695
• Precisions: [0.8257261410788381, 0.7717647058823529, 0.7255434782608695, 0.6816720257234726]
• Brevity Penalty: 0.0928
• Length Ratio: 0.2961
• Translation Length: 482
• Reference Length: 1628
• Cer: 0.7610
• Wer: 0.8275

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: 0.00015752383448484097
• 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
0.3017	1.0	253	0.7248	0.0641	[0.7525150905432596, 0.65, 0.587467362924282, 0.5276073619631901]	0.1027	0.3053	497	1628	0.7622	0.8495
0.1875	2.0	506	0.6129	0.0670	[0.7914110429447853, 0.7152777777777778, 0.6613333333333333, 0.60062893081761]	0.0974	0.3004	489	1628	0.7565	0.8375
0.1171	3.0	759	0.5027	0.0697	[0.8202479338842975, 0.7587822014051522, 0.7162162162162162, 0.6741214057507987]	0.0941	0.2973	484	1628	0.7563	0.8293
0.0432	4.0	1012	0.4983	0.0695	[0.8257261410788381, 0.7717647058823529, 0.7255434782608695, 0.6816720257234726]	0.0928	0.2961	482	1628	0.7610	0.8275

Framework versions

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