bert-base-cased-wikitext2
This model is a fine-tuned version of on the None dataset. It achieves the following results on the evaluation set:
- Loss: 6.9846
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: 24
- eval_batch_size: 24
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 7.7422 | 1.0 | 782 | 7.1373 |
| 7.0302 | 2.0 | 1564 | 6.9972 |
| 6.9788 | 3.0 | 2346 | 7.0087 |
Framework versions
- Transformers 4.11.3
- Pytorch 1.12.0+cu102
- Datasets 1.14.0
- Tokenizers 0.10.3
- Downloads last month
- 7