b4ec22af3a4d53388b0e5e27a47f01fe

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

• Loss: 0.4937
• Data Size: 1.0
• Epoch Runtime: 28.2983
• Bleu: 13.7236

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	6.3754	0	2.5825	0.0253
No log	1	437	2.3763	0.0078	2.9553	0.0026
No log	2	874	2.0762	0.0156	3.6752	0.0010
No log	3	1311	1.6510	0.0312	4.3941	0.1599
No log	4	1748	1.3430	0.0625	5.3393	0.5885
1.2511	5	2185	1.0628	0.125	7.7435	2.8029
1.0061	6	2622	0.8207	0.25	9.6885	5.6196
0.7143	7	3059	0.6326	0.5	16.2720	8.4243
0.5283	8.0	3496	0.5079	1.0	28.3000	10.3124
0.4457	9.0	3933	0.4677	1.0	28.9215	11.7349
0.3747	10.0	4370	0.4576	1.0	28.1262	12.9470
0.3296	11.0	4807	0.4456	1.0	29.6471	13.6980
0.2839	12.0	5244	0.4493	1.0	27.3556	13.2235
0.2478	13.0	5681	0.4648	1.0	28.4265	13.2207
0.2139	14.0	6118	0.4901	1.0	27.7174	13.5125
0.1942	15.0	6555	0.4937	1.0	28.2983	13.7236

Framework versions

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