mbart-de-es

This model is a fine-tuned version of facebook/mbart-large-50 on Helsinki-NLP/opus_books (https://huggingface.co/datasets/Helsinki-NLP/opus_books). It achieves the following results on the evaluation set:

• Loss: 2.4208
• Bleu: 8.1685
• Gen Len: 28.2661

Model description

This model has been delivered as part of a class project.

Intended uses & limitations

The model is aim to translating from German to Spanish. As the low BLEU score shows, this model has very low accuracy, given the limited computing resources used for its creation.

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5.6e-05
• train_batch_size: 8
• eval_batch_size: 8
• 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: 2

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
3.0574	1.0	508	2.4566	7.5927	27.4213
1.8147	2.0	1016	2.4208	8.1685	28.2661

Framework versions

• Transformers 4.51.2
• Pytorch 2.10.0+cu128
• Datasets 4.0.0
• Tokenizers 0.21.4