mbart-neutralization

This model is a fine-tuned version of facebook/mbart-large-50 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0108
• Bleu: 98.1545
• Gen Len: 18.8229

Model description

Disclaimer: this is part of a practical excerise carried out as part of the University course "Machine Traslation" of the Master's Degree in Language Processing and Applied AI to Linguistcs of Universidad de La Rioja. This model is a fine-tuned version of facebook/mbart-large-50 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0108
• Bleu: 98.1545
• Gen Len: 18.8229

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: 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
No log	1.0	440	0.0220	98.1628	18.8229
0.2273	2.0	880	0.0108	98.1545	18.8229

Framework versions

• Transformers 4.49.0
• Pytorch 2.5.1+cu124
• Datasets 3.3.2
• Tokenizers 0.21.0