335da01b9a6904f1db75ca41ee543095

This model is a fine-tuned version of facebook/mbart-large-en-ro on the Helsinki-NLP/opus_books [es-it] dataset. It achieves the following results on the evaluation set:

• Loss: 2.6391
• Data Size: 1.0
• Epoch Runtime: 191.1503
• Bleu: 5.1432

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Bleu
No log	0	0	8.0998	0	16.5454	0.3102
No log	1	721	4.5807	0.0078	18.0999	0.6153
No log	2	1442	3.7287	0.0156	20.2072	1.8996
0.0817	3	2163	3.3365	0.0312	24.2767	2.3656
0.2371	4	2884	3.0686	0.0625	29.9259	2.9688
3.1127	5	3605	2.8717	0.125	41.2808	4.0683
2.8483	6	4326	2.7351	0.25	62.0248	6.4586
2.6747	7	5047	2.5971	0.5	105.7712	5.7837
2.3817	8.0	5768	2.4642	1.0	194.6537	4.8100
2.0944	9.0	6489	2.4652	1.0	190.9924	5.0201
1.8946	10.0	7210	2.4805	1.0	193.8882	5.2146
1.6319	11.0	7931	2.5250	1.0	190.3438	5.2171
1.4676	12.0	8652	2.6391	1.0	191.1503	5.1432

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.2.0
• Tokenizers 0.22.1