8df0afd06c437608c57135b41b40984e

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

• Loss: 3.5038
• Data Size: 1.0
• Epoch Runtime: 24.0824
• Bleu: 5.1337

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	9.0327	0	2.3622	0.1623
No log	1	83	7.1015	0.0078	2.7194	0.2445
No log	2	166	4.8445	0.0156	3.7503	0.7600
No log	3	249	3.8481	0.0312	4.7475	1.3868
0.1571	4	332	3.4753	0.0625	6.1458	1.7221
0.1571	5	415	3.2803	0.125	8.3114	2.2411
0.1571	6	498	3.0909	0.25	10.6818	2.3339
0.5062	7	581	2.9228	0.5	16.0173	2.8229
2.4031	8.0	664	2.7891	1.0	27.5984	3.1924
1.9791	9.0	747	2.8202	1.0	26.3073	3.4918
1.376	10.0	830	2.9413	1.0	23.5099	4.3484
0.9997	11.0	913	3.2707	1.0	24.0464	3.9342
0.7511	12.0	996	3.5038	1.0	24.0824	5.1337

Framework versions

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