<!-- 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-base-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: 0.2887
- Precision: 0.5703
- Recall: 0.6028
- F1: 0.5861
- Accuracy: 0.9216
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 |
---|---|---|---|---|---|---|---|
0.7989 | 1.0 | 515 | 0.4610 | 0.4365 | 0.3851 | 0.4091 | 0.8867 |
0.4088 | 2.0 | 1030 | 0.3468 | 0.5133 | 0.5175 | 0.5154 | 0.9067 |
0.3144 | 3.0 | 1545 | 0.3082 | 0.5492 | 0.5532 | 0.5512 | 0.9158 |
0.2675 | 4.0 | 2060 | 0.2913 | 0.5627 | 0.5865 | 0.5744 | 0.9216 |
0.239 | 5.0 | 2575 | 0.2887 | 0.5703 | 0.6028 | 0.5861 | 0.9216 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.1+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2