gpt2_small_ZH_superbpe_65536_parallel10_42
This model was trained from scratch on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 6.3874
- Accuracy: 0.1956
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: 5e-05
- train_batch_size: 74
- eval_batch_size: 128
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|
| 8.7186 | 1.0 | 1750 | 7.8277 | 0.1474 |
| 7.5989 | 2.0 | 3500 | 7.3235 | 0.1562 |
| 7.1717 | 3.0 | 5250 | 7.0227 | 0.1644 |
| 6.8902 | 4.0 | 7000 | 6.8153 | 0.1713 |
| 6.7578 | 5.0 | 8750 | 6.6682 | 0.1788 |
| 6.5863 | 6.0 | 10500 | 6.5615 | 0.1843 |
| 6.4566 | 7.0 | 12250 | 6.4845 | 0.1893 |
| 6.3612 | 8.0 | 14000 | 6.4312 | 0.1926 |
| 6.3045 | 9.0 | 15750 | 6.3997 | 0.1946 |
| 6.2582 | 10.0 | 17500 | 6.3874 | 0.1956 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.8.0+cu128
- Datasets 4.1.1
- Tokenizers 0.19.1
