BertjeWDialDataALLQonly02
This model is a fine-tuned version of GroNLP/bert-base-dutch-cased on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.9043
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: 5e-05
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 12.0
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 2.2438 | 1.0 | 871 | 2.1122 |
| 2.1235 | 2.0 | 1742 | 2.0784 |
| 2.0712 | 3.0 | 2613 | 2.0679 |
| 2.0034 | 4.0 | 3484 | 2.0546 |
| 1.9375 | 5.0 | 4355 | 2.0277 |
| 1.8911 | 6.0 | 5226 | 2.0364 |
| 1.8454 | 7.0 | 6097 | 1.9812 |
| 1.808 | 8.0 | 6968 | 2.0175 |
| 1.7716 | 9.0 | 7839 | 2.0286 |
| 1.7519 | 10.0 | 8710 | 1.9653 |
| 1.7358 | 11.0 | 9581 | 1.9817 |
| 1.7084 | 12.0 | 10452 | 1.9633 |
Framework versions
- Transformers 4.13.0.dev0
- Pytorch 1.10.0
- Datasets 1.16.1
- Tokenizers 0.10.3