mbert-quoref
This model is a fine-tuned version of bert-base-multilingual-cased on the quoref dataset. It achieves the following results on the evaluation set:
- Loss: 3.7567
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: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.8761 | 1.0 | 1213 | 1.7360 |
| 1.305 | 2.0 | 2426 | 1.6877 |
| 0.9271 | 3.0 | 3639 | 1.8559 |
| 0.6565 | 4.0 | 4852 | 2.0420 |
| 0.4911 | 5.0 | 6065 | 2.3335 |
| 0.3468 | 6.0 | 7278 | 2.6380 |
| 0.2522 | 7.0 | 8491 | 2.8952 |
| 0.2001 | 8.0 | 9704 | 3.2514 |
| 0.1501 | 9.0 | 10917 | 3.5567 |
| 0.1314 | 10.0 | 12130 | 3.7567 |
Framework versions
- Transformers 4.34.1
- Pytorch 2.0.1+cu118
- Datasets 2.14.5
- Tokenizers 0.14.1