harouzie's picture
Update README.md
6490fc1
metadata
base_model: harouzie/mbart-translation-en2vi
tags:
  - generated_from_trainer
metrics:
  - bleu
  - sacrebleu
model-index:
  - name: mbart-translation-en2vi
    results: []
license: mit
datasets:
  - harouzie/vi_en-translation
language:
  - en
  - vi
library_name: transformers
pipeline_tag: translation

mbart-translation-en2vi

This model is a fine-tuned version of harouzie/mbart-translation-en2vi on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2959
  • Bleu: 75.2497
  • Gen Len: 12.1475

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: 2e-05
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 32
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • num_epochs: 1

Training results

Training Loss Epoch Step Validation Loss Bleu Gen Len
0.3171 1.0 6352 0.2997 74.9862 12.135

Framework versions

  • Transformers 4.33.0
  • Pytorch 2.0.0
  • Datasets 2.1.0
  • Tokenizers 0.13.3