c612ab386982ad53b25002b25eba1bbf

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [es-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 3.4808
• Data Size: 1.0
• Epoch Runtime: 13.2585
• Bleu: 11.8616

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	7.5514	0	1.3859	0.6088
No log	1	33	6.7429	0.0078	2.0347	0.6707
No log	2	66	6.2846	0.0156	3.1228	0.8001
No log	3	99	5.6446	0.0312	4.3997	1.4854
0.3229	4	132	5.1726	0.0625	6.2971	2.1596
0.3229	5	165	4.4707	0.125	7.5515	2.8881
0.3229	6	198	3.9850	0.25	10.1647	3.2474
0.7972	7	231	3.5017	0.5	11.2661	4.9605
2.1432	8.0	264	3.1175	1.0	14.4844	6.3621
2.1432	9.0	297	3.0197	1.0	13.7216	7.9238
2.3154	10.0	330	3.0554	1.0	14.2651	7.9192
1.4751	11.0	363	3.2023	1.0	14.8051	10.9879
1.4751	12.0	396	3.2886	1.0	13.1328	8.7601
0.9701	13.0	429	3.4808	1.0	13.2585	11.8616

Framework versions

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