fb9b1cbb04a9dd33c96f52c9c448047d

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

• Loss: 5.9147
• Data Size: 1.0
• Epoch Runtime: 13.9310
• Bleu: 0.6951

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	7.8892	0	1.3226	1.0609
No log	1	35	6.0862	0.0078	2.0049	3.3624
No log	2	70	4.9161	0.0156	3.3357	5.7054
No log	3	105	4.4493	0.0312	4.5513	8.7691
No log	4	140	4.2009	0.0625	5.2991	10.7303
No log	5	175	3.9092	0.125	7.6382	11.3745
No log	6	210	3.4942	0.25	9.9591	13.3354
No log	7	245	2.2325	0.5	11.4899	20.5862
0.6144	8.0	280	1.3626	1.0	15.4295	21.7102
1.2879	9.0	315	1.8147	1.0	14.7142	15.4338
9.9492	10.0	350	12.3120	1.0	15.1571	0.0
9.9492	11.0	385	7.3778	1.0	16.0151	0.0014
8.8583	12.0	420	5.9147	1.0	13.9310	0.6951

Framework versions

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