3386a0d7a2ea9dca4bfff8a276cc3ab6

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 [de-nl] dataset. It achieves the following results on the evaluation set:

• Loss: 1.8667
• Data Size: 1.0
• Epoch Runtime: 24.7443
• Bleu: 7.2120

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	7.2337	0	2.3585	0.1068
No log	1	390	4.4196	0.0078	3.4692	0.9099
No log	2	780	3.5357	0.0156	3.2514	2.1376
No log	3	1170	3.0545	0.0312	3.6427	3.0668
No log	4	1560	2.6725	0.0625	4.3328	3.9625
0.171	5	1950	2.3762	0.125	5.6561	4.6860
0.3516	6	2340	2.1345	0.25	8.2011	5.3536
2.0029	7	2730	1.9491	0.5	13.3400	6.1964
1.7362	8.0	3120	1.7969	1.0	25.0868	6.7250
1.4855	9.0	3510	1.7460	1.0	24.4173	7.0344
1.2975	10.0	3900	1.7461	1.0	25.0720	7.2060
1.1559	11.0	4290	1.7647	1.0	23.7358	7.0819
1.0103	12.0	4680	1.8180	1.0	25.7774	7.2025
0.9059	13.0	5070	1.8667	1.0	24.7443	7.2120

Framework versions

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