DALPA_GER_3kSource_3kTarget

This model is a fine-tuned version of facebook/nougat-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5235

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: 15

Training results

Training Loss	Epoch	Step	Validation Loss
8.3295	1.0	107	0.8368
5.5907	2.0	214	0.7068
4.6364	3.0	321	0.6378
3.7652	4.0	428	0.5883
3.2332	5.0	535	0.5552
3.0329	6.0	642	0.5363
3.3107	7.0	749	0.5323
3.0447	8.0	856	0.5263
2.7618	9.0	963	0.5246
2.79	10.0	1070	0.5248
2.8944	11.0	1177	0.5242
2.9454	12.0	1284	0.5235
2.9518	13.0	1391	0.5234
3.0869	14.0	1498	0.5235
2.9532	14.8652	1590	0.5235

Framework versions

• Transformers 4.47.1
• Pytorch 2.5.1+cu121
• Datasets 4.1.0
• Tokenizers 0.21.0