jind11/TextFooler
A Model for Natural Language Attack on Text Classification and Inference
This tool generates 'adversarial examples' by making tiny, human-imperceptible changes to text, which can fool AI models designed for text classification or natural language inference. It takes an existing text dataset and a trained AI model, then outputs modified text examples that trick the model into incorrect predictions. Anyone involved in evaluating or stress-testing the robustness of AI-powered text analysis systems would find this useful.
529 stars. No commits in the last 6 months.
Use this if you need to test the resilience and potential vulnerabilities of your AI text classification or natural language inference models against subtle textual manipulations.
Not ideal if you are looking to improve the general accuracy of your AI models or if your primary concern is with typical data noise rather than targeted adversarial attacks.
Stars
529
Forks
87
Language
Python
License
MIT
Category
Last pushed
Dec 08, 2022
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/jind11/TextFooler"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
thunlp/OpenAttack
An Open-Source Package for Textual Adversarial Attack.
thunlp/TAADpapers
Must-read Papers on Textual Adversarial Attack and Defense
thunlp/OpenBackdoor
An open-source toolkit for textual backdoor attack and defense (NeurIPS 2022 D&B, Spotlight)
thunlp/SememePSO-Attack
Code and data of the ACL 2020 paper "Word-level Textual Adversarial Attacking as Combinatorial...
osoleve/glitchlings
Enemies for your LLM