gpt2_tiny_zh-hk-wiki
This model was trained on a dataset with a 50MB size for 10 epochs only.
Purely intended for research and testing purposes.
This model is a fine-tuned version of jed351/gpt2-tiny-zh-hk on the cantonese-wikipedia dataset. It achieves the following results on the evaluation set:
- Loss: 3.3834
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: 200
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| No log | 1.0 | 412 | 3.6481 |
| 4.0728 | 2.0 | 824 | 3.5399 |
| 3.757 | 3.0 | 1236 | 3.4889 |
| 3.6669 | 4.0 | 1648 | 3.4557 |
| 3.6189 | 5.0 | 2060 | 3.4295 |
| 3.6189 | 6.0 | 2472 | 3.4129 |
| 3.5835 | 7.0 | 2884 | 3.3992 |
| 3.5604 | 8.0 | 3296 | 3.3905 |
| 3.5434 | 9.0 | 3708 | 3.3849 |
| 3.537 | 10.0 | 4120 | 3.3834 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.13.1+cu116
- Datasets 2.9.0
- Tokenizers 0.13.2