ddd5fb9da60b7ae478d3ac4abc978cbd

This model is a fine-tuned version of Helsinki-NLP/opus-mt-tc-bible-big-deu_eng_fra_por_spa-mul on the Helsinki-NLP/opus_books [fr-pl] dataset. It achieves the following results on the evaluation set:

• Loss: 2.3768
• Data Size: 1.0
• Epoch Runtime: 5.4013
• Bleu: 1.1587

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	8.2753	0	0.9162	0.0238
No log	1	70	6.1540	0.0078	1.3710	0.1420
No log	2	140	4.9738	0.0156	1.3461	0.1688
No log	3	210	4.0354	0.0312	1.6539	0.1348
No log	4	280	3.5166	0.0625	1.7714	0.2536
No log	5	350	3.1272	0.125	2.2604	0.5318
No log	6	420	2.7247	0.25	3.2300	0.5716
0.482	7	490	2.4319	0.5	4.3451	0.8376
2.1876	8.0	560	2.1861	1.0	6.8048	1.1013
1.8653	9.0	630	2.1413	1.0	5.5837	1.0674
1.5391	10.0	700	2.1644	1.0	5.5167	1.1017
1.3007	11.0	770	2.2021	1.0	5.5199	1.0864
1.1711	12.0	840	2.2794	1.0	5.4068	1.1810
0.9441	13.0	910	2.3768	1.0	5.4013	1.1587

Framework versions

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