metadata
language:
- en
- zh
metrics:
- sacrebleu
pipeline_tag: translation
base_model: facebook/mbart-large-cc25
eval
This model is a fine-tuned version of facebook/mbart-large-cc25 on IWSLT14 En-Zh dataset.
It achieves the following results on the evaluation set:
- eval_loss: 3.8405
- eval_bleu: 3.5173
- eval_gen_len: 21.5826
It achieves the following results on the test set:
- test_loss: 3.8337
- test_bleu: 3.277
- test_gen_len: 21.6287
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 7
- num_epochs: 9
Framework versions
- Transformers 4.35.2
- Pytorch 2.0.1+cu117
- Datasets 2.15.0
- Tokenizers 0.15.0