kd-bertBase-bertTiny
This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the glue dataset. It achieves the following results on the evaluation set:
- Loss: 1.0591
- Accuracy: 0.8268
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: 6e-05
- train_batch_size: 128
- eval_batch_size: 128
- seed: 33
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 7
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 1.4776 | 1.0 | 527 | 1.1240 | 0.8062 |
| 0.8442 | 2.0 | 1054 | 1.0475 | 0.8165 |
| 0.6568 | 3.0 | 1581 | 1.0529 | 0.8131 |
| 0.5623 | 4.0 | 2108 | 1.0503 | 0.8188 |
| 0.5066 | 5.0 | 2635 | 1.0471 | 0.8303 |
| 0.4736 | 6.0 | 3162 | 1.0711 | 0.8280 |
| 0.4603 | 7.0 | 3689 | 1.0591 | 0.8268 |
Framework versions
- Transformers 4.33.3
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.13.3