3e40a579e347c3d992adc5e8a463503b

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

• Loss: 0.7158
• Data Size: 1.0
• Epoch Runtime: 13.7145
• Bleu: 9.3971

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	7.0810	0	1.5575	0.0555
No log	1	204	4.6804	0.0078	2.7664	0.0075
No log	2	408	2.5115	0.0156	2.1550	0.0
No log	3	612	1.8473	0.0312	2.6461	0.0310
No log	4	816	1.6227	0.0625	3.1751	0.0815
No log	5	1020	1.3841	0.125	3.9490	0.1720
0.1368	6	1224	1.1728	0.25	5.3563	1.0075
1.1061	7	1428	0.9690	0.5	8.5175	3.4842
0.8266	8.0	1632	0.7642	1.0	15.3991	5.5758
0.6844	9.0	1836	0.6980	1.0	13.6537	7.3643
0.5965	10.0	2040	0.6503	1.0	13.1599	8.1337
0.5095	11.0	2244	0.6392	1.0	14.2463	8.7971
0.4307	12.0	2448	0.6409	1.0	13.6444	8.7505
0.3807	13.0	2652	0.6510	1.0	14.1114	9.2982
0.3309	14.0	2856	0.6905	1.0	13.9348	9.5636
0.2923	15.0	3060	0.7158	1.0	13.7145	9.3971

Framework versions

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