osoleve/glitchlings

Enemies for your LLM

41
/ 100
Emerging

This project offers tools to intentionally introduce realistic text corruptions into your data. You provide original text, and it outputs versions with typos, homophone substitutions, or confusable characters. It's designed for machine learning engineers and researchers who are building and testing robust language models.

Available on PyPI.

Use this if you need to test how well your language models handle real-world data imperfections or want to train models to be more resilient to errors.

Not ideal if you're looking for tools to clean or correct existing noisy text data.

natural-language-processing model-robustness data-augmentation machine-learning-engineering language-model-evaluation
Maintenance 10 / 25
Adoption 7 / 25
Maturity 24 / 25
Community 0 / 25

How are scores calculated?

Stars

35

Forks

Language

Python

License

Apache-2.0

Last pushed

Jan 20, 2026

Commits (30d)

0

Dependencies

4

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/nlp/osoleve/glitchlings"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.