<!-- 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-en3
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: 2.1864
- Bleu: 0.7037
- Gen Len: 11.0
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: 0.001
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 256
- total_train_batch_size: 2048
- 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 | 0.99 | 119 | 4.4541 | 0.0 | 5.0 |
| No log | 1.99 | 238 | 2.4214 | 0.3414 | 16.0 |
| No log | 2.99 | 357 | 2.2158 | 0.3212 | 15.0 |
| No log | 3.99 | 476 | 2.1737 | 0.3283 | 12.0 |
| 3.2958 | 4.99 | 595 | 2.1864 | 0.7037 | 11.0 |
Framework versions
- Transformers 4.18.0
- Pytorch 1.11.0+cu113
- Datasets 2.1.0
- Tokenizers 0.12.1
