tiny-bert-sst2-distilled
This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4932
- Accuracy: 0.7631
- Recall: 0.4941
- Precision: 0.7071
- F1: 0.5817
- Mcc: 0.4369
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: 0.0005029644721099001
- train_batch_size: 32
- eval_batch_size: 32
- seed: 33
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 9
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Recall | Precision | F1 | Mcc |
|---|---|---|---|---|---|---|---|---|
| 0.6337 | 1.0 | 160 | 0.6248 | 0.6667 | 0.0 | 0.0 | 0.0 | 0.0 |
| 0.5902 | 2.0 | 320 | 0.5729 | 0.6612 | 0.3318 | 0.4879 | 0.3950 | 0.1775 |
| 0.5727 | 3.0 | 480 | 0.5756 | 0.6651 | 0.4565 | 0.4974 | 0.4761 | 0.2311 |
| 0.5631 | 4.0 | 640 | 0.5504 | 0.6941 | 0.2494 | 0.5989 | 0.3522 | 0.2262 |
| 0.5486 | 5.0 | 800 | 0.5304 | 0.7176 | 0.4918 | 0.5921 | 0.5373 | 0.3396 |
| 0.5381 | 6.0 | 960 | 0.5163 | 0.7310 | 0.3976 | 0.6602 | 0.4963 | 0.3475 |
| 0.526 | 7.0 | 1120 | 0.5090 | 0.7467 | 0.5859 | 0.6288 | 0.6066 | 0.4207 |
| 0.5149 | 8.0 | 1280 | 0.4971 | 0.7584 | 0.5176 | 0.6811 | 0.5882 | 0.4297 |
| 0.5088 | 9.0 | 1440 | 0.4932 | 0.7631 | 0.4941 | 0.7071 | 0.5817 | 0.4369 |
Framework versions
- Transformers 4.40.0
- Pytorch 2.2.1+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1