<!-- 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. -->
xlm-roberta-base-es-aug-ner
This model is a fine-tuned version of xlm-roberta-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.2611
- Precision: 0.6234
- Recall: 0.6361
- F1: 0.6297
- Accuracy: 0.9279
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: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 5
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
0.6289 | 1.0 | 827 | 0.3766 | 0.4739 | 0.4639 | 0.4688 | 0.8968 |
0.2458 | 2.0 | 1654 | 0.2911 | 0.5828 | 0.5573 | 0.5698 | 0.9202 |
0.2043 | 3.0 | 2481 | 0.2739 | 0.6110 | 0.5971 | 0.6039 | 0.9256 |
0.1662 | 4.0 | 3308 | 0.2617 | 0.6265 | 0.6336 | 0.6300 | 0.9282 |
0.146 | 5.0 | 4135 | 0.2611 | 0.6234 | 0.6361 | 0.6297 | 0.9279 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2