Team Ai
Modelpublic

ctrlbuzz/bert-addresses

sourceHugging Faceupdated 3y agoView on Hugging Face
9likes1kdownloads
Model Card

Model Card for Model ID

This model is developed to tag Names, Organisations and addresses. I have used a data combined fro Conll, ontonotes5, and a custom address dataset that was self made. Cleaned out the tags. Detects U.S addresses. [\"O\", \"B-ORG\", \"I-ORG\", \"B-PER\", \"I-PER\",'B-addr','I-addr']

Model Description

  • —Developed by: ctrlbuzz
  • —Model type: Bert
  • —Language(s) (NLP): Named Entity recognition
  • —Finetuned from model [optional]: bert-base-cased

Uses

Direct Use

python
from transformers import AutoTokenizer, AutoModelForTokenClassification
from transformers import pipeline
tokenizer = AutoTokenizer.from_pretrained('bert-base-cased')
model = AutoModelForTokenClassification.from_pretrained("ctrlbuzz/bert-addresses")
nlp = pipeline("ner", model=model, tokenizer=tokenizer)
example = "While Maria was representing Johnson & Associates at a conference in Spain, she mailed me a letter from her new office at 123 Elm St., Apt. 4B, Springfield, IL.",

print(nlp(example))