This model is a fine-tuned version of Helsinki-NLP/opus-mt-de-en with the dataset wmt16 for the pair of languages german-english. A tutorial for this task is available in the files.

Intended uses & limitations

Limitations: scholar use.

Training and evaluation data

Training done with keras from Transformers. Evaluation with Bleu score.

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

optimizer: {'name': 'AdamWeightDecay', 'learning_rate': {'class_name': 'PolynomialDecay', 'config': {'initial_learning_rate': 5e-05, 'decay_steps': 1245, 'end_learning_rate': 0.0, 'power': 1.0, 'cycle': False, 'name': None}}, 'decay': 0.0, 'beta_1': 0.9, 'beta_2': 0.999, 'epsilon': 1e-08, 'amsgrad': False, 'weight_decay_rate': 0.01}
training_precision: mixed_float16

Training results

Train Loss	Validation Loss	Epoch
1.5115	1.4061	0
1.2931	1.4111	1
1.1590	1.4200	2
1.0644	1.4324	3
1.0077	1.4381	4

Framework versions

Transformers 4.27.4
TensorFlow 2.12.0
Datasets 2.11.0
Tokenizers 0.13.2