350370eb933919f125d2e53ccc15b60d

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

• Loss: 4.0890
• Data Size: 1.0
• Epoch Runtime: 27.0292
• Bleu: 5.4253

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	9.7490	0	2.4550	0.0878
No log	1	89	8.5883	0.0078	2.8691	0.0892
No log	2	178	7.6396	0.0156	4.5832	0.1557
No log	3	267	7.0854	0.0312	5.8591	0.4334
No log	4	356	6.3964	0.0625	7.9275	0.8445
No log	5	445	5.6296	0.125	9.2789	1.0920
0.416	6	534	4.7496	0.25	11.5732	1.6098
1.603	7	623	4.1200	0.5	16.4060	2.0472
3.5632	8.0	712	3.6963	1.0	28.8332	3.4359
2.8546	9.0	801	3.5499	1.0	29.2812	3.7016
2.3679	10.0	890	3.5325	1.0	27.1797	4.1455
1.8597	11.0	979	3.6358	1.0	27.3711	4.1458
1.4479	12.0	1068	3.7857	1.0	26.8330	4.9092
1.1463	13.0	1157	3.9284	1.0	27.1676	5.7018
0.8843	14.0	1246	4.0890	1.0	27.0292	5.4253

Framework versions

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