distilbert_add_GLUE_Experiment_logit_kd_mnli_96
This model is a fine-tuned version of distilbert-base-uncased on the GLUE MNLI dataset. It achieves the following results on the evaluation set:
- Loss: 0.5576
- Accuracy: 0.5239
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 | Accuracy |
|---|---|---|---|---|
| 0.624 | 1.0 | 1534 | 0.6178 | 0.3605 |
| 0.6176 | 2.0 | 3068 | 0.6138 | 0.3767 |
| 0.6139 | 3.0 | 4602 | 0.6112 | 0.3822 |
| 0.6104 | 4.0 | 6136 | 0.6071 | 0.3977 |
| 0.6027 | 5.0 | 7670 | 0.5978 | 0.4091 |
| 0.5958 | 6.0 | 9204 | 0.6104 | 0.4151 |
| 0.5877 | 7.0 | 10738 | 0.5963 | 0.4517 |
| 0.5787 | 8.0 | 12272 | 0.6054 | 0.4627 |
| 0.5711 | 9.0 | 13806 | 0.5753 | 0.4905 |
| 0.5641 | 10.0 | 15340 | 0.5713 | 0.4987 |
| 0.5583 | 11.0 | 16874 | 0.5645 | 0.5115 |
| 0.5535 | 12.0 | 18408 | 0.5646 | 0.5117 |
| 0.549 | 13.0 | 19942 | 0.5692 | 0.5176 |
| 0.5456 | 14.0 | 21476 | 0.5613 | 0.5220 |
| 0.5425 | 15.0 | 23010 | 0.5584 | 0.5302 |
| 0.5399 | 16.0 | 24544 | 0.5641 | 0.5252 |
| 0.5375 | 17.0 | 26078 | 0.5628 | 0.5260 |
| 0.5353 | 18.0 | 27612 | 0.5659 | 0.5200 |
| 0.533 | 19.0 | 29146 | 0.5676 | 0.5310 |
| 0.5311 | 20.0 | 30680 | 0.5563 | 0.5323 |
| 0.5291 | 21.0 | 32214 | 0.5682 | 0.5250 |
| 0.5274 | 22.0 | 33748 | 0.5661 | 0.5282 |
| 0.5255 | 23.0 | 35282 | 0.5673 | 0.5325 |
| 0.5236 | 24.0 | 36816 | 0.5563 | 0.5416 |
| 0.5219 | 25.0 | 38350 | 0.5703 | 0.5290 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2