tatoeba-tok-ru

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-ru on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4694
• Bleu: 20.4311

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 15
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu
2.0538	1.0	1167	1.7944	7.2757
1.7461	2.0	2334	1.6562	14.4046
1.5644	3.0	3501	1.5849	15.8651
1.4611	4.0	4668	1.5486	16.9808
1.3698	5.0	5835	1.5209	15.9727
1.305	6.0	7002	1.5049	18.1714
1.2458	7.0	8169	1.4947	18.7180
1.2006	8.0	9336	1.4868	19.5074
1.1555	9.0	10503	1.4800	19.5799
1.1251	10.0	11670	1.4759	20.0295
1.091	11.0	12837	1.4741	20.1306
1.0701	12.0	14004	1.4710	19.7831
1.0433	13.0	15171	1.4715	20.5833
1.0359	14.0	16338	1.4694	20.3306
1.0263	15.0	17505	1.4702	20.3269

Framework versions

• Transformers 4.52.4
• Pytorch 2.7.1+cu128
• Datasets 3.6.0
• Tokenizers 0.21.1