tiny-vanilla-target-glue-mnli
This model is a fine-tuned version of google/bert_uncased_L-2_H-128_A-2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.8100
- Accuracy: 0.6375
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 | Accuracy |
|---|---|---|---|---|
| 1.0866 | 0.04 | 500 | 1.0515 | 0.4557 |
| 1.0101 | 0.08 | 1000 | 0.9526 | 0.5612 |
| 0.9599 | 0.12 | 1500 | 0.9195 | 0.5802 |
| 0.9378 | 0.16 | 2000 | 0.9018 | 0.5930 |
| 0.9229 | 0.2 | 2500 | 0.8904 | 0.5954 |
| 0.9182 | 0.24 | 3000 | 0.8802 | 0.6033 |
| 0.9019 | 0.29 | 3500 | 0.8738 | 0.6070 |
| 0.8971 | 0.33 | 4000 | 0.8613 | 0.6154 |
| 0.8788 | 0.37 | 4500 | 0.8593 | 0.6172 |
| 0.8856 | 0.41 | 5000 | 0.8508 | 0.6194 |
| 0.8751 | 0.45 | 5500 | 0.8404 | 0.6256 |
| 0.8718 | 0.49 | 6000 | 0.8445 | 0.6248 |
| 0.8739 | 0.53 | 6500 | 0.8333 | 0.6306 |
| 0.8653 | 0.57 | 7000 | 0.8363 | 0.6280 |
| 0.8588 | 0.61 | 7500 | 0.8213 | 0.6376 |
| 0.8587 | 0.65 | 8000 | 0.8215 | 0.6360 |
| 0.8544 | 0.69 | 8500 | 0.8268 | 0.6292 |
| 0.8556 | 0.73 | 9000 | 0.8045 | 0.6463 |
| 0.8445 | 0.77 | 9500 | 0.8187 | 0.6328 |
| 0.836 | 0.81 | 10000 | 0.8021 | 0.6446 |
| 0.8399 | 0.86 | 10500 | 0.8100 | 0.6375 |
Framework versions
- Transformers 4.26.0.dev0
- Pytorch 1.13.0+cu116
- Datasets 2.8.1.dev0
- Tokenizers 0.13.2