---
language: es
license: cc-by-nc-sa-4.0
tags:
  - ner
  - token-classification
  - bilstm
  - crf
  - conll2002
pipeline_tag: token-classification
datasets:
  - conll2002
metrics:
  - f1
  - precision
  - recall
---

# bilstm-crf-fasttext NER CoNLL-2002 (bs=16)

BiLSTM + CRF con embeddings **FastText español** (cc.es.300).
Entrenado sobre CoNLL-2002 español. Modelo CUSTOM con `trust_remote_code=True`.

## Métricas en test (CoNLL-2002)

| Metric | Valor |
|:------:|:-----:|
| F1 | **0.8079** |
| Precision | 0.8129 |
| Recall | 0.8030 |

## Uso

```python
# Requiere: pip install pytorch-crf
from transformers import AutoModelForTokenClassification, AutoConfig
import json

config = AutoConfig.from_pretrained("cvalenciaunivalle/bilstm-crf-fasttext-conll-bs16", trust_remote_code=True)
model  = AutoModelForTokenClassification.from_pretrained("cvalenciaunivalle/bilstm-crf-fasttext-conll-bs16", trust_remote_code=True)

# Cargar vocab (incluye word2idx, char2idx, id2tag)
from huggingface_hub import hf_hub_download
vocab_path = hf_hub_download("cvalenciaunivalle/bilstm-crf-fasttext-conll-bs16", "vocab.json")
with open(vocab_path) as f:
    vocab = json.load(f)

# Predecir
oraciones_tokenizadas = [["Juan", "vive", "en", "Bogotá", "."]]
tags = model.predict(oraciones_tokenizadas, vocab)
print(tags)  # [['B-PER', 'O', 'O', 'B-LOC', 'O']]
```

## Detalles

- Arquitectura: BiLSTM hidden=200, dropout=0.3, CRF de salida
- Word embeddings: 300d (FastText)

- Batch size: 16, épocas: 10
- Compañeros: Laura Valero, Cristian Valencia
- Taller PLN — Maestría Univalle