ab3e83f04214a5b915fbcdfd0bce2d98

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

• Loss: 3.3780
• Data Size: 1.0
• Epoch Runtime: 24.7403
• Bleu: 3.2583

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	10.1248	0	2.6815	0.1495
No log	1	83	7.9142	0.0078	3.1837	0.1096
No log	2	166	6.5836	0.0156	3.8360	0.2253
No log	3	249	5.2607	0.0312	4.7108	0.3079
0.2033	4	332	4.2376	0.0625	6.3747	0.9628
0.2033	5	415	3.6341	0.125	8.4884	1.5667
0.2033	6	498	3.2302	0.25	10.7253	2.1015
0.6024	7	581	3.0042	0.5	15.4209	2.6434
2.6274	8.0	664	2.8218	1.0	26.1005	3.0005
2.2001	9.0	747	2.8343	1.0	26.5943	3.4102
1.6375	10.0	830	2.9799	1.0	24.0733	3.3775
1.3023	11.0	913	3.1830	1.0	24.5920	3.3252
1.0581	12.0	996	3.3780	1.0	24.7403	3.2583

Framework versions

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