GLTR (Giant Language Model Test Room)
gltr.io
Spec sheet
- Purpose
- GLTR is a tool to help people assess whether text was generated by a language model.Read 5 Oct 2026 · arxiv.org
- Audience
- The paper describes its detection methods as simple and explainable to non-experts.Read 5 Oct 2026 · arxiv.org
- License
- The repository identifies the project license as Apache 2.Read 5 Oct 2026 · github.com
- Founded
- 2019Read 4 Oct 2026 · gltr.io
- Availability
- The project paper describes GLTR as open-source and publicly deployed.Read 5 Oct 2026 · arxiv.org
- Customization
- Developers can add a custom model by implementing a backend API that inherits from AbstractLanguageChecker and registering it.Read 5 Oct 2026 · github.com
- Detection methods
- It applies baseline statistical methods to identify generation artifacts across common sampling schemes.Read 5 Oct 2026 · arxiv.org
- Models
- The project’s server supports gpt-2-small, BERT, or a custom model.Read 5 Oct 2026 · github.com
- Self-hosting
- The repository documents installing dependencies and running the server locally.Read 5 Oct 2026 · github.com
- Study result
- In a human-subjects study, GLTR’s annotation scheme raised fake-text detection from 54% to 72% without prior training.Read 5 Oct 2026 · arxiv.org
- Web demo
- The repository links to an online demo at gltr.io/dist/index.html.Read 5 Oct 2026 · github.com
Where Specifiction places it
- AI Content Detection Software#42 of 456.6