opus-mt-lg-en-informal-v1
This model is a fine-tuned version of Helsinki-NLP/opus-mt-lg-en on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.1343
- Bleu: 0.0585
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
- lr_scheduler_warmup_steps: 500
- num_epochs: 20
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Bleu |
|---|---|---|---|---|
| 4.568 | 1.0 | 119 | 0.8525 | 0.0314 |
| 0.6495 | 2.0 | 238 | 0.1701 | 0.0487 |
| 0.1889 | 3.0 | 357 | 0.1443 | 0.0761 |
| 0.1513 | 4.0 | 476 | 0.1384 | 0.0587 |
| 0.1424 | 5.0 | 595 | 0.1357 | 0.0790 |
| 0.1331 | 6.0 | 714 | 0.1346 | 0.0502 |
| 0.1265 | 7.0 | 833 | 0.1340 | 0.0961 |
| 0.1296 | 8.0 | 952 | 0.1339 | 0.0763 |
| 0.123 | 9.0 | 1071 | 0.1340 | 0.0800 |
| 0.1227 | 10.0 | 1190 | 0.1339 | 0.0428 |
| 0.117 | 11.0 | 1309 | 0.1343 | 0.0585 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.1.2
- Datasets 2.19.1
- Tokenizers 0.19.1