Repo

janus

Calibrated, non-autoregressive tool-call failure prediction via single-token masking. 175 tests, honest negative results on R-Judge and Agent-SafetyBench.

Visit github.com ↗

Built with

Python

Comments · 0

No comments yet. Tried it? Tell Arihant what worked, what didn't, or what you'd want next.

More from members

All projects

See what other members are building and what each project is looking for.