aa19f74aee421fa56464908c2ff1914f

This model is a fine-tuned version of facebook/mbart-large-cc25 on the Helsinki-NLP/opus_books [fr-it] dataset. It achieves the following results on the evaluation set:

• Loss: 2.6385
• Data Size: 1.0
• Epoch Runtime: 100.4334
• Bleu: 6.5399

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	10.1497	0	9.1597	0.1302
No log	1	367	3.8361	0.0078	10.9271	3.3716
No log	2	734	3.4969	0.0156	12.0964	3.5905
No log	3	1101	3.2710	0.0312	14.5424	4.2795
No log	4	1468	3.0594	0.0625	17.3586	5.0796
0.1808	5	1835	2.7732	0.125	23.0156	5.7441
2.8014	6	2202	2.5462	0.25	34.5393	6.9994
2.4527	7	2569	2.3890	0.5	56.5874	7.5023
2.1601	8.0	2936	2.2911	1.0	101.6380	6.3681
1.8687	9.0	3303	2.2917	1.0	100.7508	6.7366
1.5353	10.0	3670	2.3731	1.0	100.5286	6.5660
1.2721	11.0	4037	2.5060	1.0	100.9613	6.4634
1.0636	12.0	4404	2.6385	1.0	100.4334	6.5399

Framework versions

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