82a4525fe3173240f6b0eeb778e123ee

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [es-no] dataset. It achieves the following results on the evaluation set:

• Loss: 4.1403
• Data Size: 1.0
• Epoch Runtime: 26.1926
• Bleu: 6.2203

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	8.8344	0	2.4197	0.1307
No log	1	89	8.0141	0.0078	3.4174	0.1667
No log	2	178	7.3740	0.0156	4.6124	0.2295
No log	3	267	6.8809	0.0312	6.1710	0.4788
No log	4	356	6.1093	0.0625	8.1774	0.6554
No log	5	445	5.5263	0.125	9.9323	0.8811
0.401	6	534	4.8542	0.25	11.8390	1.5366
1.6257	7	623	4.2970	0.5	16.8881	2.2534
3.7355	8.0	712	3.8367	1.0	28.8878	4.1167
2.9653	9.0	801	3.6534	1.0	29.5484	4.9744
2.4375	10.0	890	3.6643	1.0	25.9042	7.5595
1.9203	11.0	979	3.7713	1.0	26.1067	6.3694
1.4857	12.0	1068	3.9128	1.0	27.4656	7.4301
1.1865	13.0	1157	4.1403	1.0	26.1926	6.2203

Framework versions

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