results

This model is a fine-tuned version of Helsinki-NLP/opus-mt-de-en on the wmt16 dataset. It achieves the following results on the evaluation set:

• Loss: 1.3984
• Bleu-1: 0.8788
• Bleu-2: 0.8122
• Rouge-l: 0.6251

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu-1	Bleu-2	Rouge-l
2.2098	1.0	995	2.1014	0.8223	0.7409	0.5191
1.7011	2.0	1990	2.1251	0.8195	0.7382	0.517
1.3455	3.0	2985	2.1688	0.8204	0.7387	0.5154
1.1516	4.0	3980	2.2028	0.8201	0.7383	0.5151

Framework versions

• Transformers 4.27.4
• Pytorch 2.0.0+cu118
• Datasets 2.11.0
• Tokenizers 0.13.2