e839463c636388ec25751c13a39fc409

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

• Loss: 2.7795
• Data Size: 1.0
• Epoch Runtime: 182.6636
• Bleu: 5.2160

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	7.0464	0	15.3566	0.3402
No log	1	721	3.7530	0.0078	16.9556	1.9147
No log	2	1442	3.3320	0.0156	19.8584	2.6015
0.0679	3	2163	3.1126	0.0312	23.9248	3.5278
0.2152	4	2884	2.9380	0.0625	28.6742	4.8340
2.8904	5	3605	2.7901	0.125	41.1273	4.7562
2.6575	6	4326	2.6388	0.25	59.8350	4.7446
2.4937	7	5047	2.5062	0.5	101.5607	5.3234
2.1773	8.0	5768	2.3840	1.0	185.4566	5.8490
1.8506	9.0	6489	2.4062	1.0	183.4434	5.8160
1.6132	10.0	7210	2.4809	1.0	182.0601	5.4004
1.3236	11.0	7931	2.6044	1.0	183.2010	5.5577
1.133	12.0	8652	2.7795	1.0	182.6636	5.2160

Framework versions

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