1fa86ab49633eca7aca10397b60ffff0

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [es-ru] dataset. It achieves the following results on the evaluation set:

• Loss: 2.5765
• Data Size: 1.0
• Epoch Runtime: 113.2025
• Bleu: 6.4224

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	13.2378	0	10.0300	0.1952
No log	1	419	4.3222	0.0078	11.1390	2.4396
No log	2	838	4.0372	0.0156	12.2807	3.2403
0.1316	3	1257	3.6789	0.0312	14.4593	3.5790
0.1316	4	1676	3.0180	0.0625	19.0017	3.9439
0.2101	5	2095	2.5260	0.125	25.2641	4.3364
0.338	6	2514	2.3748	0.25	37.3263	4.6741
2.2244	7	2933	2.2532	0.5	62.0737	5.0513
1.9575	8.0	3352	2.1841	1.0	112.9038	5.7303
1.7514	9.0	3771	2.2680	1.0	110.9497	6.9838
1.478	10.0	4190	2.2870	1.0	111.3968	7.8966
1.1858	11.0	4609	2.3975	1.0	112.5540	8.7753
0.9776	12.0	5028	2.5765	1.0	113.2025	6.4224

Framework versions

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