c655a4934f3ca10767e53c0dd70183d4

This model is a fine-tuned version of facebook/mbart-large-50-many-to-one-mmt on the Helsinki-NLP/opus_books [es-ru] dataset. It achieves the following results on the evaluation set:

• Loss: 2.7025
• Data Size: 1.0
• Epoch Runtime: 106.3718
• Bleu: 5.9885

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	6.4716	0	9.3555	0.5167
No log	1	419	5.0980	0.0078	10.3912	0.9845
No log	2	838	4.2275	0.0156	11.7913	1.7460
0.1249	3	1257	3.7453	0.0312	13.8284	2.3493
0.1249	4	1676	3.4165	0.0625	17.2913	2.8509
0.2196	5	2095	3.1088	0.125	23.5053	3.4322
0.3956	6	2514	2.8318	0.25	34.9478	4.2485
2.501	7	2933	2.5790	0.5	58.7882	4.8576
2.0719	8.0	3352	2.3813	1.0	108.7493	5.2926
1.7032	9.0	3771	2.3542	1.0	107.5828	5.7128
1.3802	10.0	4190	2.3984	1.0	106.6468	5.7494
1.0278	11.0	4609	2.5031	1.0	106.3281	5.8263
0.8138	12.0	5028	2.6012	1.0	106.7620	5.6944
0.5794	13.0	5447	2.7025	1.0	106.3718	5.9885

Framework versions

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