a28f2cdd2ca81f74fa4189e51bc2a573

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

• Loss: 1.6652
• Data Size: 1.0
• Epoch Runtime: 13.0733
• Bleu: 35.6936

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	5.2877	0	1.3414	4.4799
No log	1	35	3.4020	0.0078	1.8052	9.3876
No log	2	70	2.4452	0.0156	3.4253	20.3433
No log	3	105	2.2423	0.0312	4.6224	21.0147
No log	4	140	2.0691	0.0625	5.7155	19.3118
No log	5	175	1.9055	0.125	7.6372	16.6085
No log	6	210	1.6750	0.25	10.4390	21.5632
No log	7	245	1.3032	0.5	10.7081	28.6433
0.3148	8.0	280	1.2410	1.0	14.1789	37.2413
0.8051	9.0	315	1.3000	1.0	13.5443	40.7802
0.3813	10.0	350	1.4309	1.0	13.7919	41.2118
0.3813	11.0	385	1.5523	1.0	14.6303	36.4598
0.1544	12.0	420	1.6652	1.0	13.0733	35.6936

Framework versions

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