metadata
license: cc-by-nc-4.0
tags:
- generated_from_trainer
metrics:
- bleu
model-index:
- name: hi_do_epoch_5
results: []
hi_do_epoch_5
This model is a fine-tuned version of facebook/nllb-200-distilled-600M on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.4675
- Bleu: 34.8763
- Gen Len: 40.6362
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: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu | Gen Len |
|---|---|---|---|---|---|
| No log | 1.0 | 398 | 2.0556 | 22.5137 | 39.5758 |
| 2.9556 | 2.0 | 796 | 1.6863 | 30.1421 | 39.9786 |
| 1.8135 | 3.0 | 1194 | 1.5487 | 32.8717 | 40.1963 |
| 1.5932 | 4.0 | 1592 | 1.4839 | 34.2659 | 40.5695 |
| 1.5932 | 5.0 | 1990 | 1.4675 | 34.8763 | 40.6362 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.10.1+cu113
- Datasets 2.8.0
- Tokenizers 0.13.2