distilbert_sa_GLUE_Experiment_logit_kd_data_aug_stsb_256
This model is a fine-tuned version of distilbert-base-uncased on the GLUE STSB dataset. It achieves the following results on the evaluation set:
- Loss: 1.4500
- Pearson: 0.1761
- Spearmanr: 0.1778
- Combined Score: 0.1770
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: 256
- eval_batch_size: 256
- seed: 10
- distributed_type: multi-GPU
- 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 | Pearson | Spearmanr | Combined Score |
|---|---|---|---|---|---|---|
| 0.5832 | 1.0 | 1259 | 1.5244 | 0.1737 | 0.1803 | 0.1770 |
| 0.2202 | 2.0 | 2518 | 1.4500 | 0.1761 | 0.1778 | 0.1770 |
| 0.1249 | 3.0 | 3777 | 1.4720 | 0.1743 | 0.1782 | 0.1762 |
| 0.0822 | 4.0 | 5036 | 1.5790 | 0.1581 | 0.1658 | 0.1619 |
| 0.0611 | 5.0 | 6295 | 1.4750 | 0.1850 | 0.1905 | 0.1878 |
| 0.0477 | 6.0 | 7554 | 1.5776 | 0.1612 | 0.1694 | 0.1653 |
| 0.0394 | 7.0 | 8813 | 1.5512 | 0.1648 | 0.1694 | 0.1671 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2