f5d4fed39a6c669bce14ef54d4cd44b9

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

• Loss: 2.5398
• Data Size: 1.0
• Epoch Runtime: 25.1672
• Bleu: 20.9207

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	7.1686	0	2.2277	0.6095
No log	1	77	5.7081	0.0078	2.9007	2.7157
No log	2	154	4.8581	0.0156	4.3699	4.2872
No log	3	231	4.5930	0.0312	6.3694	5.5668
No log	4	308	4.3610	0.0625	7.6610	5.8156
No log	5	385	4.0103	0.125	10.5672	6.6641
0.3762	6	462	2.8249	0.25	12.1878	7.6777
1.1157	7	539	2.3501	0.5	14.8541	13.3462
1.8487	8.0	616	2.0610	1.0	25.2958	18.3477
1.3867	9.0	693	2.0915	1.0	24.1034	16.7896
0.7946	10.0	770	2.2663	1.0	23.9708	17.7514
0.651	11.0	847	2.4562	1.0	24.6602	21.2667
0.3676	12.0	924	2.5398	1.0	25.1672	20.9207

Framework versions

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