7c3eaf6a26effdb8e2ec526e1936801e

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 [es-ru] dataset. It achieves the following results on the evaluation set:

• Loss: 0.5968
• Data Size: 1.0
• Epoch Runtime: 26.9685
• Bleu: 7.8341

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	6.8374	0	2.4865	0.0612
No log	1	419	2.3447	0.0078	3.9790	0.0012
No log	2	838	1.8985	0.0156	3.2234	0.0189
0.0664	3	1257	1.6371	0.0312	3.9522	0.0715
0.0664	4	1676	1.3557	0.0625	4.6337	0.2346
0.091	5	2095	1.1510	0.125	7.2401	1.1013
0.1404	6	2514	0.9300	0.25	9.4978	2.7620
0.811	7	2933	0.7430	0.5	15.3156	4.4534
0.6405	8.0	3352	0.6122	1.0	27.3244	6.2251
0.5502	9.0	3771	0.5731	1.0	27.0987	6.8139
0.4889	10.0	4190	0.5510	1.0	26.4232	8.1712
0.4279	11.0	4609	0.5415	1.0	26.6588	7.8366
0.3805	12.0	5028	0.5570	1.0	26.6086	7.7537
0.329	13.0	5447	0.5578	1.0	26.4377	8.4514
0.3202	14.0	5866	0.5712	1.0	26.5208	8.3533
0.2703	15.0	6285	0.5968	1.0	26.9685	7.8341

Framework versions

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