bert-tiny-emotion-KD-BERT
This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the emotion dataset. It achieves the following results on the evaluation set:
- Loss: 0.4810
- Accuracy: 0.9175
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: 16
- eval_batch_size: 16
- seed: 33
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 50
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 3.8247 | 1.0 | 1000 | 2.5170 | 0.7745 |
| 1.9864 | 2.0 | 2000 | 1.3436 | 0.874 |
| 1.1126 | 3.0 | 3000 | 0.8299 | 0.894 |
| 0.6924 | 4.0 | 4000 | 0.6500 | 0.9025 |
| 0.5272 | 5.0 | 5000 | 0.6097 | 0.908 |
| 0.4298 | 6.0 | 6000 | 0.5913 | 0.904 |
| 0.3936 | 7.0 | 7000 | 0.5165 | 0.9135 |
| 0.3238 | 8.0 | 8000 | 0.5120 | 0.9075 |
| 0.3018 | 9.0 | 9000 | 0.4989 | 0.916 |
| 0.2605 | 10.0 | 10000 | 0.4810 | 0.9175 |
| 0.2512 | 11.0 | 11000 | 0.4757 | 0.9135 |
| 0.219 | 12.0 | 12000 | 0.4676 | 0.914 |
| 0.2046 | 13.0 | 13000 | 0.4794 | 0.911 |
Framework versions
- Transformers 4.22.1
- Pytorch 1.12.1+cu113
- Datasets 2.5.1
- Tokenizers 0.12.1