<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
opus-mt-ko-en-finetuned-ko-to-en5
This model is a fine-tuned version of Helsinki-NLP/opus-mt-ko-en on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.1434
- Bleu: 52.6052
- Gen Len: 8.1982
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 256
- total_train_batch_size: 1024
- 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 | 105 | 1.8436 | 35.225 | 8.1735 |
| No log | 2.0 | 210 | 1.4106 | 44.7159 | 8.1923 |
| No log | 3.0 | 315 | 1.2410 | 49.5117 | 8.2165 |
| No log | 4.0 | 420 | 1.1661 | 51.8883 | 8.201 |
| 1.8123 | 5.0 | 525 | 1.1434 | 52.6052 | 8.1982 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.11.0+cu113
- Datasets 2.1.0
- Tokenizers 0.12.1
