Kullback logo

Kullback

Synthetic RL Environments from Traces

Share on:
Listing image

Kullback is Leibler

Leibler · Kullback Skip to content leibler .dev How it works Kullback GitHub kullback is the open-source harness behind Leibler. Your traces in. An environment out. Traces are the logs your agent already writes. Kullback reads them, rebuilds the tools, data and rules the agent used, and replays the logs to check the rebuild. The claim we are working toward, not made yet: a 2B parameter model post-trained in that environment performs on the real tasks. GitHub Design philosophy new traces, fixes, disputes traces your runs build tools, state, checks run any model verdict code, end state report yo

Related listings

Check if a PDF's redactions can be undone, in the browser

Kerning Game for Designers

A biotech catalyst calendar that links every date to the SEC sentence

catches the fake source links AI search engines invent

pixel-art loops for each of Claude Code's 188 spinner verbs

animated illustrations and skills to make more like them