opus-mt-fr-de-new-finetuned-fr-to-wol

This model is a fine-tuned version of Helsinki-NLP/opus-mt-fr-de on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.4897
• Bleu: 5.5073
• Gen Len: 61.5873

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: 0.0003
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	91	3.0424	0.4073	288.4603
No log	2.0	182	2.6751	0.9893	166.5666
No log	3.0	273	2.4795	3.1938	80.8081
No log	4.0	364	2.3826	4.1209	73.1684
No log	5.0	455	2.3629	4.858	63.6266
2.5384	6.0	546	2.3635	5.1207	66.588
2.5384	7.0	637	2.3929	5.3246	61.1974
2.5384	8.0	728	2.4382	5.5606	63.0014
2.5384	9.0	819	2.4625	5.3472	62.4086
2.5384	10.0	910	2.4897	5.5073	61.5873

Framework versions

• Transformers 4.38.1
• Pytorch 2.1.2
• Datasets 2.1.0
• Tokenizers 0.15.2