Sebastian Raschka open-sources an end-to-end 'AI text detector from scratch' project
rasbt · x · 2026-09-20
Sebastian Raschka released the full codebase accompanying his article 'Build an AI Text Detector From Scratch'. The project covers the entire pipeline: three-stage dataset construction (licensed human text, AI-generated text validated with Pangram, AI-assisted text), comparisons of several classifier architectures, and experiments using the detector as a verifier during reinforcement learning. The binary classifier scores text 0–100 for likelihood of being AI-generated; he stresses the score reflects the classifier and its training distribution, not proof of authorship. Numbered scripts, tests, and MIT license make it directly reproducible.
More from Research
- Have we seen an acceleration in discoveries? Cyber spikes, math rises, algorithms flat — soumitrashukla9 · 2026-09-20
- TovanaEngine: a local world model trained on 50K+ real SWE-bench coding-agent runs — Decent-Ad9950 · 2026-09-20
- After Chess and Go: Can Any AI Engine Actually Beat Humans at Scrabble? — zuilserip · 2026-09-20
- Textbook author: 99.9% accuracy can mean zero scientific discoveries — bravo_abad · 2026-09-20
- rasbt: Jev's Secret Sauce Is Data, Not the Algorithm — Laya Rival Falls Far Short — RichmanRonald · 2026-09-20
- Sebastian Raschka: restricting LLM outputs to an action space is just a classic encoder classifier — rasbt · 2026-09-20