6c6b198dc81e61faa7e9317ed6ed7161

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

• Loss: 3.3106
• Data Size: 1.0
• Epoch Runtime: 23.6074
• Bleu: 8.1137

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	7.1275	0	2.3096	0.2920
No log	1	77	6.7173	0.0078	3.0325	0.3157
No log	2	154	6.0862	0.0156	4.6845	0.5176
No log	3	231	5.5384	0.0312	6.3120	0.6620
No log	4	308	5.0979	0.0625	7.9874	1.1160
No log	5	385	4.5694	0.125	10.6663	1.9943
0.4467	6	462	4.0594	0.25	12.0912	2.9261
1.6017	7	539	3.5878	0.5	14.6060	4.1821
3.0105	8.0	616	3.1255	1.0	24.6607	5.9233
2.4785	9.0	693	2.9708	1.0	24.0123	8.7416
1.7189	10.0	770	2.9837	1.0	23.2241	7.6218
1.4736	11.0	847	3.0294	1.0	23.6414	9.0353
0.9628	12.0	924	3.0967	1.0	24.5745	8.5747
0.7135	13.0	1001	3.3106	1.0	23.6074	8.1137

Framework versions

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