Edit model card

mbart-neutralization

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

  • Loss: 0.0181
  • Bleu: 98.7341
  • Gen Len: 18.4896

Model description

Used and trained as part of the Master's Degree in NLP of Universidad de la Rioja 2023-2024

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: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 2

Training results

Training Loss Epoch Step Validation Loss Bleu Gen Len
No log 1.0 440 0.0307 91.2911 18.25
0.2343 2.0 880 0.0181 98.7341 18.4896

Framework versions

  • Transformers 4.37.2
  • Pytorch 2.1.0+cu121
  • Datasets 2.17.1
  • Tokenizers 0.15.2
Downloads last month
2
Safetensors
Model size
611M params
Tensor type
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for Tinaenhugging/mbart-neutralization

Finetuned
(128)
this model