_Nougat_Ger_01

This model is a fine-tuned version of facebook/nougat-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.3141

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: 0.0001
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 6
• total_train_batch_size: 48
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss
7.8593	1.0	84	1.3509
6.4407	2.0	168	1.2018
5.3764	3.0	252	1.1687
4.8124	4.0	336	1.1438
4.2572	5.0	420	1.1438
4.0467	6.0	504	1.1760
3.4963	7.0	588	1.1957
3.0197	8.0	672	1.2227
2.7683	9.0	756	1.2486
2.3057	10.0	840	1.2624
2.486	11.0	924	1.2799
2.0421	12.0	1008	1.2846
2.0073	13.0	1092	1.2923
1.9756	14.0	1176	1.3006
1.8417	15.0	1260	1.3083
1.8013	16.0	1344	1.3110
1.6453	17.0	1428	1.3132
1.7862	18.0	1512	1.3141
1.7968	19.0	1596	1.3137
1.7583	19.768	1660	1.3141

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 3.3.0
• Tokenizers 0.21.0