171c57196dc48c47d6b033e3b65ae08a

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

• Loss: 5.1176
• Data Size: 1.0
• Epoch Runtime: 23.5840
• Bleu: 1.5798

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	10.0411	0	2.4258	0.1590
No log	1	70	8.7733	0.0078	2.7538	0.1818
No log	2	140	7.8017	0.0156	4.3505	0.1014
No log	3	210	7.1187	0.0312	5.5276	0.1907
No log	4	280	6.6710	0.0625	7.4885	0.2558
No log	5	350	6.1894	0.125	9.3672	0.4256
No log	6	420	5.6262	0.25	12.1285	0.9007
0.9188	7	490	5.0453	0.5	14.6722	0.8441
4.7109	8.0	560	4.6352	1.0	24.0995	1.0578
4.227	9.0	630	4.5180	1.0	23.4284	1.2043
3.6523	10.0	700	4.4951	1.0	24.6473	1.4657
3.1624	11.0	770	4.5555	1.0	22.4846	1.3114
2.8972	12.0	840	4.7185	1.0	23.6383	1.3867
2.3231	13.0	910	4.8316	1.0	23.8681	1.5662
2.0318	14.0	980	5.1176	1.0	23.5840	1.5798

Framework versions

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