<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
bert-base-spanish-wwm-uncased-finetuned-squad_es
This model is a fine-tuned version of dccuchile/bert-base-spanish-wwm-uncased on the squad_es dataset. It achieves the following results on the evaluation set:
- Loss: 1.7747
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: 3
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.5377 | 1.0 | 8259 | 1.4632 |
| 1.1928 | 2.0 | 16518 | 1.5536 |
| 0.9486 | 3.0 | 24777 | 1.7747 |
Framework versions
- Transformers 4.19.2
- Pytorch 1.11.0+cu113
- Datasets 2.2.2
- Tokenizers 0.12.1
