small-mlm-glue-wnli-custom-tokenizer
This model is a fine-tuned version of google/bert_uncased_L-4_H-512_A-8 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2966
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: constant
- num_epochs: 200
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 5.7189 | 6.25 | 500 | 5.1827 |
| 4.4287 | 12.5 | 1000 | 4.3325 |
| 3.5065 | 18.75 | 1500 | 3.3465 |
| 2.6482 | 25.0 | 2000 | 2.4543 |
| 1.8682 | 31.25 | 2500 | 1.8391 |
| 1.405 | 37.5 | 3000 | 1.3834 |
| 0.9971 | 43.75 | 3500 | 1.0132 |
| 0.7485 | 50.0 | 4000 | 0.8970 |
| 0.5753 | 56.25 | 4500 | 0.4850 |
| 0.4756 | 62.5 | 5000 | 0.4589 |
| 0.3926 | 68.75 | 5500 | 0.7330 |
| 0.3287 | 75.0 | 6000 | 0.3611 |
| 0.2929 | 81.25 | 6500 | 0.3802 |
| 0.2477 | 87.5 | 7000 | 0.2649 |
| 0.2316 | 93.75 | 7500 | 0.3877 |
| 0.2207 | 100.0 | 8000 | 0.1159 |
| 0.2015 | 106.25 | 8500 | 0.2502 |
| 0.1803 | 112.5 | 9000 | 0.1365 |
| 0.1544 | 118.75 | 9500 | 0.2966 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu116
- Datasets 2.8.1.dev0
- Tokenizers 0.13.2