103ba3ea440804a8bbd6a627f72eef7a

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

• Loss: 3.1177
• Data Size: 1.0
• Epoch Runtime: 22.1512
• Bleu: 8.7676

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	9.1584	0	2.1950	0.1742
No log	1	74	5.9016	0.0078	3.0285	0.8436
No log	2	148	3.9611	0.0156	4.3972	3.3328
0.187	3	222	3.3275	0.0312	5.6988	4.4431
0.187	4	296	3.1228	0.0625	7.5820	4.8098
0.215	5	370	2.9899	0.125	9.7266	5.4777
0.215	6	444	2.8418	0.25	11.8515	6.2551
0.5618	7	518	2.7125	0.5	14.6921	9.6673
1.4748	8.0	592	2.5998	1.0	24.4487	7.7326
1.6008	9.0	666	2.6680	1.0	23.7369	6.6063
1.2934	10.0	740	2.7845	1.0	22.2259	6.8376
0.822	11.0	814	2.9672	1.0	22.1685	6.9700
0.6293	12.0	888	3.1177	1.0	22.1512	8.7676

Framework versions

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