Instructions to use nationaldesignstudio/rampart with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers.js
How to use nationaldesignstudio/rampart with Transformers.js:
// npm i @huggingface/transformers import { pipeline } from '@huggingface/transformers'; // Allocate pipeline const pipe = await pipeline('token-classification', 'nationaldesignstudio/rampart');
Uncommon hotel and building names are not censored out
I modified one of the examples to include a hotel name with a room number:
Hi, I'm Sarah Johnson. You can reach me at sarah.johnson@gmail.com or 212-867-5309. My SSN is 472-81-0094 and I live at 1234 Maple Street, Springfield, IL 62704, but I am currently on vacation in Hawaii and I am currently staying at the Hilton hotel, my room number is 203.
The model correctly edits out 203 as a BUILDING_NUMBER, which I guess is used for unit numbers for apartment buildings. But the "Hilton hotel" should be redacted out since street names are also edited out - a hotel name points to a specific building while a street name usually points to a group of buildings.
I actually just noticed that if you use a common street name as the hotel's name, such as "Rosetti Hotel", it would correctly edit out "Rosetti", but not "hotel".