QData/TextAttack
A Python framework for running adversarial attacks on NLP models, augmenting training data, and training NLP models.
Not currently ranked — collecting fresh signals.
star history
TextAttack provides researchers and practitioners with tools to evaluate NLP model robustness by running adversarial attacks that test model vulnerabilities. It includes a library of attack algorithms and transformation techniques for generating adversarial examples. The framework also supports data augmentation to improve model generalization and offers command-line utilities for training NLP models.
Frequently asked
- What is QData/TextAttack?
- A Python framework for running adversarial attacks on NLP models, augmenting training data, and training NLP models.
- Is TextAttack open source?
- Yes — QData/TextAttack is open source, released under the MIT license.
- What language is TextAttack written in?
- QData/TextAttack is primarily written in Python.
- How popular is TextAttack?
- QData/TextAttack has 3.5k stars on GitHub.
- Where can I find TextAttack?
- QData/TextAttack is on GitHub at https://github.com/QData/TextAttack.