bert_large_subjqa_model_v3
This model is a fine-tuned version of bert-large-uncased-whole-word-masking-finetuned-squad on the None dataset. It achieves the following results on the evaluation set:
- Loss: 8.8423
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: 2e-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: 15
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| No log | 1.0 | 410 | 2.3971 |
| 2.5672 | 2.0 | 820 | 2.7380 |
| 1.6385 | 3.0 | 1230 | 3.1387 |
| 0.8467 | 4.0 | 1640 | 4.0478 |
| 0.4177 | 5.0 | 2050 | 4.9824 |
| 0.4177 | 6.0 | 2460 | 5.5992 |
| 0.2208 | 7.0 | 2870 | 6.2417 |
| 0.1469 | 8.0 | 3280 | 7.1139 |
| 0.0966 | 9.0 | 3690 | 8.0166 |
| 0.0746 | 10.0 | 4100 | 7.7937 |
| 0.0589 | 11.0 | 4510 | 7.8738 |
| 0.0589 | 12.0 | 4920 | 8.3362 |
| 0.0439 | 13.0 | 5330 | 8.6546 |
| 0.0388 | 14.0 | 5740 | 8.8050 |
| 0.029 | 15.0 | 6150 | 8.8423 |
Framework versions
- Transformers 4.28.0
- Pytorch 1.13.0a0+d321be6
- Datasets 2.12.0
- Tokenizers 0.13.3