distilbert-base-uncased__sst2__train-16-0
This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.6903
- Accuracy: 0.5091
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: 4
- eval_batch_size: 4
- seed: 42
- 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 |
|---|---|---|---|---|
| 0.6934 | 1.0 | 7 | 0.7142 | 0.2857 |
| 0.6703 | 2.0 | 14 | 0.7379 | 0.2857 |
| 0.6282 | 3.0 | 21 | 0.7769 | 0.2857 |
| 0.5193 | 4.0 | 28 | 0.8799 | 0.2857 |
| 0.5104 | 5.0 | 35 | 0.8380 | 0.4286 |
| 0.2504 | 6.0 | 42 | 0.8622 | 0.4286 |
| 0.1794 | 7.0 | 49 | 0.9227 | 0.4286 |
| 0.1156 | 8.0 | 56 | 0.8479 | 0.4286 |
| 0.0709 | 9.0 | 63 | 1.0929 | 0.2857 |
| 0.0471 | 10.0 | 70 | 1.2189 | 0.2857 |
| 0.0288 | 11.0 | 77 | 1.2026 | 0.4286 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2
- Tokenizers 0.10.3