f72674be1a7ebf39863484da66dd3b08

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

• Loss: 5.2315
• Data Size: 1.0
• Epoch Runtime: 22.4934
• Bleu: 1.3656

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	10.0610	0	2.4389	0.1732
No log	1	70	8.8490	0.0078	3.1240	0.2108
No log	2	140	8.0251	0.0156	4.3762	0.1183
No log	3	210	7.1935	0.0312	6.4993	0.3605
No log	4	280	6.7176	0.0625	7.8125	0.3968
No log	5	350	6.2144	0.125	9.3547	0.6097
No log	6	420	5.6163	0.25	11.2288	0.3876
0.9221	7	490	5.0689	0.5	13.6342	0.7422
4.7508	8.0	560	4.6926	1.0	23.0617	0.9754
4.2715	9.0	630	4.8511	1.0	23.3665	0.3748
3.8964	10.0	700	4.5051	1.0	21.9111	1.0717
3.394	11.0	770	4.4952	1.0	22.0606	1.1542
3.1352	12.0	840	4.5971	1.0	22.0392	1.2154
2.5401	13.0	910	4.8014	1.0	21.6713	1.2601
2.2014	14.0	980	5.0182	1.0	22.6577	1.2034
1.8292	15.0	1050	5.2315	1.0	22.4934	1.3656

Framework versions

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