e67c6ae68fe9cb18e7dbb369134417f6

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

• Loss: 2.9249
• Data Size: 1.0
• Epoch Runtime: 28.5944
• Bleu: 5.1300

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	7.1845	0	3.1091	0.2522
No log	1	88	5.8304	0.0078	3.7539	0.3160
No log	2	176	4.6177	0.0156	4.9868	0.4891
No log	3	264	3.9173	0.0312	6.9181	0.6316
No log	4	352	3.2978	0.0625	8.5433	1.8424
No log	5	440	2.9429	0.125	10.6738	2.7452
0.2744	6	528	2.6733	0.25	13.4026	3.1128
0.9237	7	616	2.5383	0.5	16.8254	3.7549
2.2743	8.0	704	2.4183	1.0	28.9946	4.2885
1.8286	9.0	792	2.4790	1.0	29.2875	5.1088
1.4105	10.0	880	2.5391	1.0	25.4701	4.9466
1.1189	11.0	968	2.6972	1.0	26.5608	4.6423
0.8715	12.0	1056	2.9249	1.0	28.5944	5.1300

Framework versions

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