distilbert_sst2_finetuned
This model is a fine-tuned version of distilbert-base-uncased on the glue dataset. It achieves the following results on the evaluation set:
- Loss: 0.2831
- Accuracy: 0.875
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: 1e-06
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 4
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 0.6883 | 0.24 | 500 | 0.6768 | 0.5115 |
| 0.5422 | 0.48 | 1000 | 0.4060 | 0.8200 |
| 0.3479 | 0.71 | 1500 | 0.3533 | 0.8452 |
| 0.3217 | 0.95 | 2000 | 0.3343 | 0.8567 |
| 0.2967 | 1.19 | 2500 | 0.3200 | 0.8635 |
| 0.2857 | 1.43 | 3000 | 0.3110 | 0.8624 |
| 0.2723 | 1.66 | 3500 | 0.3010 | 0.8670 |
| 0.2744 | 1.9 | 4000 | 0.2896 | 0.8727 |
| 0.2594 | 2.14 | 4500 | 0.2897 | 0.8716 |
| 0.2574 | 2.38 | 5000 | 0.2845 | 0.8761 |
| 0.2484 | 2.61 | 5500 | 0.2869 | 0.8739 |
| 0.2464 | 2.85 | 6000 | 0.2842 | 0.8761 |
| 0.2451 | 3.09 | 6500 | 0.2820 | 0.8773 |
| 0.2504 | 3.33 | 7000 | 0.2805 | 0.8784 |
| 0.236 | 3.56 | 7500 | 0.2833 | 0.875 |
| 0.2366 | 3.8 | 8000 | 0.2831 | 0.875 |
Framework versions
- Transformers 4.27.4
- Pytorch 1.13.1+cu116
- Datasets 2.11.0
- Tokenizers 0.13.2