language:
- es
license: apache-2.0
datasets:
- eriktks/conll2002
metrics:
- precision
- recall
- f1
- accuracy
pipeline_tag: token-classification
Model Name: bert-finetuned-ner-1
This is a BERT model fine-tuned for Named Entity Recognition (NER).
Model Description
This is a fine-tuned BERT model for Named Entity Recognition (NER) task using CONLL2002 dataset.
In the first part, the dataset must be pre-processed in order to give it to the model. This is done using the 🤗 Transformers and BERT tokenizers. Once this is done, finetuning is applied from bert-base-cased and using the 🤗 AutoModelForTokenClassification.
Finally, the model is trained obtaining the neccesary metrics for evaluating its performance (Precision, Recall, F1 and Accuracy)
Summary of executed tests can be found in: https://docs.google.com/spreadsheets/d/1lI7skNIvRurwq3LA5ps7JFK5TxToEx4s7Kaah3ezyQc/edit?usp=sharing
Model can be found in: https://huggingface.co./paulrojasg/bert-finetuned-ner-1
Github repository: https://github.com/paulrojasg/nlp_4th_workshop
Training
Training Details
- Epochs: 10
- Learning Rate: 2e-05
- Weight Decay: 0.01
- Batch Size (Train): 16
- Batch Size (Eval): 8
Training Metrics
| Epoch | Training Loss | Validation Loss | Precision | Recall | F1 Score | Accuracy |
|---|---|---|---|---|---|---|
| 1 | 0.1729 | 0.1462 | 0.6739 | 0.7376 | 0.7043 | 0.9590 |
| 2 | 0.0750 | 0.1432 | 0.7030 | 0.7684 | 0.7342 | 0.9625 |
| 3 | 0.0496 | 0.1394 | 0.7725 | 0.7983 | 0.7852 | 0.9667 |
| 4 | 0.0351 | 0.1460 | 0.7678 | 0.8017 | 0.7844 | 0.9672 |
| 5 | 0.0255 | 0.1521 | 0.7656 | 0.8081 | 0.7863 | 0.9678 |
| 6 | 0.0182 | 0.1709 | 0.7573 | 0.8015 | 0.7787 | 0.9667 |
| 7 | 0.0134 | 0.1753 | 0.7794 | 0.8159 | 0.7973 | 0.9691 |
| 8 | 0.0110 | 0.1806 | 0.7674 | 0.8125 | 0.7893 | 0.9685 |
| 9 | 0.0080 | 0.1938 | 0.7724 | 0.8139 | 0.7926 | 0.9683 |
| 10 | 0.0072 | 0.1938 | 0.7722 | 0.8139 | 0.7925 | 0.9685 |
Authors
Made by:
- Paul Rodrigo Rojas Guerrero
- Jose Luis Hincapie Bucheli
- Sebastián Idrobo Avirama
With help from: