f86898e1d74480c2e40704479a7cc112

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

• Loss: 2.4257
• Data Size: 1.0
• Epoch Runtime: 24.9307
• Bleu: 13.1544

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	4.7419	0	2.3672	4.8796
No log	1	77	3.4240	0.0078	2.9990	6.6236
No log	2	154	3.0114	0.0156	4.4623	6.6648
No log	3	231	2.5882	0.0312	6.5263	6.9898
No log	4	308	2.4208	0.0625	7.5569	8.2677
No log	5	385	2.2941	0.125	10.5918	8.8518
0.2219	6	462	2.1743	0.25	11.8573	9.2941
0.8199	7	539	2.0734	0.5	14.7191	12.2275
1.5496	8.0	616	2.0091	1.0	25.3823	13.6524
1.2045	9.0	693	2.0449	1.0	24.3005	13.9529
0.6844	10.0	770	2.1916	1.0	23.6840	13.6916
0.5514	11.0	847	2.3205	1.0	23.9082	13.5624
0.2721	12.0	924	2.4257	1.0	24.9307	13.1544

Framework versions

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