auto-improve

GAN-style self-improvement loop for any text artifact: mutate, grade with a SEPARATE model, keep only verified wins (pairwise-judged), revert the rest. The git

What is auto-improve?

GAN-style self-improvement loop for any text artifact: mutate, grade with a SEPARATE model, keep only verified wins (pairwise-judged), revert the rest. The git history is the improvement log.

Add this badge to your README

Featured on YunoDB Badge styles & help

Maintain this project? Show that it is featured on YunoDB.

Markdown[![Featured on YunoDB](https://yunodb.com/badge/crimeacsauto-improve.svg)](https://yunodb.com/tools/crimeacsauto-improve?utm_source=github&utm_medium=badge)
HTML<a href="https://yunodb.com/tools/crimeacsauto-improve?utm_source=github&utm_medium=badge"><img src="https://yunodb.com/badge/crimeacsauto-improve.svg" alt="Featured on YunoDB" height="20"></a>