sdocio's picture
Update README.md: info about dataset is pending
dfcb43b
|
raw
history blame
1.92 kB
---
language: es
license: gpl-3.0
tags:
- PyTorch
- Transformers
- Token Classification
- roberta
- roberta-base-bne
widget:
- text: "Fue antes de llegar a Sigüeiro, en el Camino de Santiago."
- text: "El proyecto lo financia el Ministerio de Industria y Competitividad."
model-index:
- name: roberta-bne-ner-cds
results: []
---
# Introduction
This model is a fine-tuned version of [roberta-base-bne](https://huggingface.co/PlanTL-GOB-ES/roberta-base-bne) for Named-Entity Recognition, in the domain of tourism related to the Way of Saint Jacques. It recognizes four types of entities: location (LOC), organizations (ORG), person (PER) and miscellaneous (MISC).
## Usage
You can use this model with Transformers *pipeline* for NER.
```python
from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
tokenizer = AutoTokenizer.from_pretrained("roberta-bne-ner-cds")
model = AutoModelForTokenClassification.from_pretrained("roberta-bne-ner-cds")
example = "Fue antes de llegar a Sigüeiro, en el Camino de Santiago. El proyecto lo financia el Ministerio de Industria y Competitividad."
ner_pipe = pipeline('ner', model=model, tokenizer=tokenizer, aggregation_strategy="simple")
for ent in ner_pipe(example):
print(ent)
```
## Dataset
ToDo
## Model performance
entity|precision|recall|f1
-|-|-|-
PER|0.965|0.924|0.944
ORG|0.900|0.701|0.788
LOC|0.982|0.985|0.983
MISC|0.798|0.874|0.834
micro avg|0.964|0.968|0.966|4265
macro avg|0.911|0.871|0.887
weighted avg|0.965|0.968|0.966
## Training procedure
### Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3.0
### Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2