dea384ffe57b783ba91d88bd96d6647e

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

• Loss: 3.6685
• Data Size: 1.0
• Epoch Runtime: 25.6428
• Bleu: 6.2692

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	9.5591	0	2.5580	0.1238
No log	1	89	8.0775	0.0078	3.3588	0.1408
No log	2	178	7.0613	0.0156	4.7479	0.5632
No log	3	267	6.4729	0.0312	6.0752	0.7620
No log	4	356	5.9320	0.0625	7.9954	1.0448
No log	5	445	5.0818	0.125	9.4720	1.6021
0.3825	6	534	4.3272	0.25	11.5219	2.3738
1.4449	7	623	3.8296	0.5	16.5534	2.7803
3.2253	8.0	712	3.4625	1.0	27.9601	3.8951
2.4747	9.0	801	3.3183	1.0	27.8773	4.0499
1.961	10.0	890	3.3388	1.0	26.4003	4.3528
1.4651	11.0	979	3.4519	1.0	26.5708	4.6929
1.068	12.0	1068	3.5578	1.0	27.8804	5.6759
0.8189	13.0	1157	3.6685	1.0	25.6428	6.2692

Framework versions

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