tatoeba-tok-fr

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

• Loss: 1.3544
• Bleu: 23.4181

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
1.8608	1.0	1167	1.6433	13.4398
1.593	2.0	2334	1.5195	17.7658
1.4354	3.0	3501	1.4565	15.5366
1.3449	4.0	4668	1.4217	19.8814
1.2681	5.0	5835	1.4016	19.6534
1.2046	6.0	7002	1.3853	21.0562
1.1545	7.0	8169	1.3738	19.8606
1.1186	8.0	9336	1.3694	20.6401
1.0803	9.0	10503	1.3615	20.3338
1.0504	10.0	11670	1.3601	23.0924
1.021	11.0	12837	1.3570	22.4592
1.0041	12.0	14004	1.3547	22.5261
0.9865	13.0	15171	1.3546	23.0413
0.9706	14.0	16338	1.3544	23.1603
0.9658	15.0	17505	1.3548	23.7041

Framework versions

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