3bbb100fd387ff30f96bf4cef38a11ba

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

• Loss: 4.4173
• Data Size: 1.0
• Epoch Runtime: 23.0477
• Bleu: 4.3792

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	10.5689	0	2.6073	0.1005
No log	1	74	8.7752	0.0078	3.1437	0.1063
No log	2	148	7.7012	0.0156	4.4864	0.1659
0.2851	3	222	6.8951	0.0312	5.6485	0.2368
0.2851	4	296	6.4349	0.0625	7.6766	0.3462
0.4401	5	370	5.9469	0.125	9.8241	0.4445
0.4401	6	444	5.4212	0.25	11.9965	0.9168
1.1679	7	518	4.8935	0.5	15.2185	1.2423
2.9891	8.0	592	4.2895	1.0	25.2073	2.2190
3.6127	9.0	666	4.0812	1.0	24.4445	2.9642
3.1862	10.0	740	3.9781	1.0	23.2731	3.7125
2.5387	11.0	814	3.9915	1.0	22.8062	4.1490
2.2161	12.0	888	4.0888	1.0	23.1630	4.0714
1.7062	13.0	962	4.2233	1.0	24.0383	4.3048
1.4444	14.0	1036	4.4173	1.0	23.0477	4.3792

Framework versions

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