xml-roberta-large-finetuned-ner
Este es modelo resultado de un finetuning de FacebookAI/xlm-roberta-large-finetuned-conll03-english sobre el conll2002 dataset. Los siguientes son los resultados sobre el conjunto de evaluación:
- Loss: 0.1364
- Precision: 0.8806
- Recall: 0.8897
- F1: 0.8851
- Accuracy: 0.9806
Model description
Este es el modelo más grande de roberta FacebookAI/xlm-roberta-large-finetuned-conll03-english- Este modelo fue ajustado usando el framework Kaggle [https://www.kaggle.com/settings]. Para realizar el preentrenamiento del modelo se tuvo que crear un directorio temporal en Kaggle con el fin de almacenar de manera temoporal el modelo que pesa alrededor de 35 Gz.
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: 4
- eval_batch_size: 8
- 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 |
---|---|---|---|---|---|---|---|
0.0743 | 1.0 | 2081 | 0.1131 | 0.8385 | 0.8587 | 0.8485 | 0.9771 |
0.049 | 2.0 | 4162 | 0.1429 | 0.8492 | 0.8564 | 0.8528 | 0.9756 |
0.031 | 3.0 | 6243 | 0.1298 | 0.8758 | 0.8817 | 0.8787 | 0.9800 |
0.0185 | 4.0 | 8324 | 0.1279 | 0.8827 | 0.8890 | 0.8859 | 0.9808 |
0.0125 | 5.0 | 10405 | 0.1364 | 0.8806 | 0.8897 | 0.8851 | 0.9806 |
Framework versions
- Transformers 4.41.1
- Pytorch 2.1.2
- Datasets 2.19.1
- Tokenizers 0.19.1
- Downloads last month
- 65
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.
Model tree for raulgdp/xml-roberta-large-finetuned-ner
Dataset used to train raulgdp/xml-roberta-large-finetuned-ner
Evaluation results
- Precision on conll2002validation set self-reported0.881
- Recall on conll2002validation set self-reported0.890
- F1 on conll2002validation set self-reported0.885
- Accuracy on conll2002validation set self-reported0.981