bert-base-uncased-pretrained-mlm-coqa-stories
This model was trained from scratch on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.8310
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: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.0573 | 1.0 | 2479 | 1.8805 |
| 1.9517 | 2.0 | 4958 | 1.8377 |
| 1.9048 | 3.0 | 7437 | 1.8310 |
Framework versions
- Transformers 4.15.0
- Pytorch 1.10.0+cu111
- Datasets 1.17.0
- Tokenizers 0.10.3