distilbert-base-uncased__sst2__train-8-1
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: 0.6930
- Accuracy: 0.5047
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 |
|---|---|---|---|---|
| 0.7082 | 1.0 | 3 | 0.7048 | 0.25 |
| 0.6761 | 2.0 | 6 | 0.7249 | 0.25 |
| 0.6653 | 3.0 | 9 | 0.7423 | 0.25 |
| 0.6212 | 4.0 | 12 | 0.7727 | 0.25 |
| 0.5932 | 5.0 | 15 | 0.8098 | 0.25 |
| 0.5427 | 6.0 | 18 | 0.8496 | 0.25 |
| 0.5146 | 7.0 | 21 | 0.8992 | 0.25 |
| 0.4356 | 8.0 | 24 | 0.9494 | 0.25 |
| 0.4275 | 9.0 | 27 | 0.9694 | 0.25 |
| 0.3351 | 10.0 | 30 | 0.9968 | 0.25 |
| 0.2812 | 11.0 | 33 | 1.0056 | 0.5 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.2+cu102
- Datasets 1.18.2
- Tokenizers 0.10.3