XLNet-Reddit-Toxic-Comment-Classification-10-epochs
This model is a fine-tuned version of xlnet-base-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.3201
- Rmse: 0.2815
- Accuracy: 0.9208
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: 3e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss | Rmse | Accuracy |
|---|---|---|---|---|---|
| 0.373 | 1.0 | 1073 | 0.3241 | 0.3053 | 0.9068 |
| 0.2543 | 2.0 | 2146 | 0.3201 | 0.2815 | 0.9208 |
| 0.2 | 3.0 | 3219 | 0.4140 | 0.2764 | 0.9236 |
| 0.1553 | 4.0 | 4292 | 0.3502 | 0.2896 | 0.9161 |
| 0.1273 | 5.0 | 5365 | 0.4866 | 0.3007 | 0.9096 |
| 0.0914 | 6.0 | 6438 | 0.4858 | 0.2781 | 0.9226 |
| 0.0808 | 7.0 | 7511 | 0.4164 | 0.2713 | 0.9264 |
| 0.0555 | 8.0 | 8584 | 0.4323 | 0.2731 | 0.9254 |
| 0.0448 | 9.0 | 9657 | 0.4782 | 0.2847 | 0.9189 |
| 0.0339 | 10.0 | 10730 | 0.5052 | 0.2864 | 0.9180 |
Framework versions
- Transformers 4.35.0.dev0
- Pytorch 2.0.0
- Datasets 2.1.0
- Tokenizers 0.14.1