3e6f04a7f580fff4e77927be189c7ead

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

• Loss: 3.8829
• Data Size: 1.0
• Epoch Runtime: 13.6613
• Bleu: 7.0048

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	9.1123	0	1.8556	0.3877
No log	1	33	7.7458	0.0078	2.1867	0.4919
No log	2	66	7.6973	0.0156	3.5352	0.6163
No log	3	99	6.6116	0.0312	4.4278	0.7818
0.3837	4	132	6.0142	0.0625	6.4574	1.0570
0.3837	5	165	5.3947	0.125	8.1821	1.8165
0.3837	6	198	4.7677	0.25	9.4717	2.6966
0.9557	7	231	4.0618	0.5	11.6033	4.0616
2.5859	8.0	264	3.5142	1.0	15.4673	5.3237
2.5859	9.0	297	3.3195	1.0	14.3756	5.9305
2.7899	10.0	330	3.2773	1.0	15.2793	6.3899
1.9164	11.0	363	3.3919	1.0	15.6622	6.7028
1.9164	12.0	396	3.4975	1.0	16.6538	7.7383
1.3737	13.0	429	3.7263	1.0	13.0813	7.7900
0.9253	14.0	462	3.8829	1.0	13.6613	7.0048

Framework versions

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