mobilebert_add_GLUE_Experiment_logit_kd_qqp_128
This model is a fine-tuned version of google/mobilebert-uncased on the GLUE QQP dataset. It achieves the following results on the evaluation set:
- Loss: 0.7254
- Accuracy: 0.7763
- F1: 0.6592
- Combined Score: 0.7178
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: 128
- eval_batch_size: 128
- 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
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Combined Score |
|---|---|---|---|---|---|---|
| 1.1514 | 1.0 | 2843 | 0.9015 | 0.7377 | 0.6339 | 0.6858 |
| 0.8815 | 2.0 | 5686 | 0.8232 | 0.7538 | 0.6095 | 0.6817 |
| 0.8373 | 3.0 | 8529 | 0.8122 | 0.7591 | 0.6325 | 0.6958 |
| 0.8086 | 4.0 | 11372 | 0.8008 | 0.7562 | 0.6018 | 0.6790 |
| 0.7833 | 5.0 | 14215 | 0.7891 | 0.7638 | 0.6390 | 0.7014 |
| 0.7568 | 6.0 | 17058 | 0.7867 | 0.7629 | 0.6188 | 0.6908 |
| 0.7227 | 7.0 | 19901 | 0.7667 | 0.7717 | 0.6623 | 0.7170 |
| 0.6868 | 8.0 | 22744 | 0.7315 | 0.7760 | 0.6597 | 0.7179 |
| 0.6563 | 9.0 | 25587 | 0.7254 | 0.7763 | 0.6592 | 0.7178 |
| 0.6325 | 10.0 | 28430 | 0.7326 | 0.7768 | 0.6775 | 0.7272 |
| 0.6116 | 11.0 | 31273 | 0.7327 | 0.7795 | 0.6748 | 0.7272 |
| 0.59 | 12.0 | 34116 | 0.7386 | 0.7813 | 0.6779 | 0.7296 |
| 0.5703 | 13.0 | 36959 | 0.7522 | 0.7806 | 0.6776 | 0.7291 |
| 0.5516 | 14.0 | 39802 | 0.7574 | 0.7776 | 0.7031 | 0.7403 |
Framework versions
- Transformers 4.26.0
- Pytorch 1.14.0a0+410ce96
- Datasets 2.9.0
- Tokenizers 0.13.2