e4c96c4d9c99b5fa58fd8014b30921a1

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-pt] dataset. It achieves the following results on the evaluation set:

• Loss: 3.1791
• Data Size: 1.0
• Epoch Runtime: 4.4008
• Bleu: 5.5316

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	7.8412	0	0.7343	0.0519
No log	1	31	6.2580	0.0078	0.8732	0.4088
No log	2	62	5.4015	0.0156	1.2009	1.4253
No log	3	93	5.1687	0.0312	1.4948	1.6174
No log	4	124	4.9171	0.0625	1.8112	2.0558
No log	5	155	4.5832	0.125	2.2033	2.2806
No log	6	186	4.1533	0.25	2.6924	2.7255
0.6994	7	217	3.5708	0.5	3.1008	3.5578
0.6994	8.0	248	3.1293	1.0	4.4502	4.5396
2.1196	9.0	279	2.9898	1.0	4.5210	4.9174
2.135	10.0	310	2.9844	1.0	4.7501	4.9846
2.135	11.0	341	2.9762	1.0	4.1724	5.0833
1.5661	12.0	372	3.0106	1.0	3.6730	5.0880
1.12	13.0	403	3.0680	1.0	3.5568	5.5468
1.12	14.0	434	3.1215	1.0	4.3660	5.6562
0.8147	15.0	465	3.1791	1.0	4.4008	5.5316

Framework versions

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