tatoeba-fr-tok

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

• Loss: 0.6225
• Bleu: 47.8803

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
1.1114	1.0	1167	0.8982	37.0049
0.8626	2.0	2334	0.7635	42.3598
0.7411	3.0	3501	0.7142	43.3202
0.6872	4.0	4668	0.6826	45.2163
0.6436	5.0	5835	0.6596	46.2536
0.6095	6.0	7002	0.6495	47.2469
0.5828	7.0	8169	0.6398	44.2553
0.5626	8.0	9336	0.6323	46.5712
0.542	9.0	10503	0.6306	47.7902
0.528	10.0	11670	0.6264	46.9619
0.512	11.0	12837	0.6247	47.1636
0.5036	12.0	14004	0.6226	47.9356
0.4924	13.0	15171	0.6232	47.5674
0.4848	14.0	16338	0.6226	47.6125
0.4824	15.0	17505	0.6225	47.5300

Framework versions

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