distilbert_add_GLUE_Experiment_logit_kd_mnli_256
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: 0.5429
- Accuracy: 0.5632
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 |
|---|---|---|---|---|
| 0.6159 | 1.0 | 1534 | 0.5957 | 0.4585 |
| 0.5816 | 2.0 | 3068 | 0.5643 | 0.5034 |
| 0.5632 | 3.0 | 4602 | 0.5612 | 0.5090 |
| 0.5542 | 4.0 | 6136 | 0.5529 | 0.5238 |
| 0.5478 | 5.0 | 7670 | 0.5510 | 0.5363 |
| 0.5424 | 6.0 | 9204 | 0.5511 | 0.5222 |
| 0.5376 | 7.0 | 10738 | 0.5523 | 0.5268 |
| 0.5328 | 8.0 | 12272 | 0.5537 | 0.5278 |
| 0.5267 | 9.0 | 13806 | 0.5482 | 0.5508 |
| 0.5187 | 10.0 | 15340 | 0.5505 | 0.5422 |
| 0.5096 | 11.0 | 16874 | 0.5406 | 0.5647 |
| 0.4998 | 12.0 | 18408 | 0.5438 | 0.5625 |
| 0.4893 | 13.0 | 19942 | 0.5415 | 0.5765 |
| 0.4793 | 14.0 | 21476 | 0.5417 | 0.5762 |
| 0.4698 | 15.0 | 23010 | 0.5425 | 0.5796 |
| 0.4609 | 16.0 | 24544 | 0.5497 | 0.5824 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2