Non-autoregressive System 1 decision engine. Typed choice, score and yes/no decisions over any text in a single forward pass, in 100+ languages, with a router that picks the right checkpoint per request.
Python calibrationclassificationdecision-modelhuggingfacejevmodernbertmultilingualnlppythonpytorchroutingtyped-decisionszero-shot
docs(crewai): say what threshold means for a score question
abhijithneilabrahamde3c5696ca1b
diff --git a/docs/crewai.md b/docs/crewai.mdindex da66f11f..fa1403d0 100644--- a/docs/crewai.md+++ b/docs/crewai.md@@ -97,6 +97,8 @@ except LayaTaskGuardError as e:print(f"Blocked by LayaTaskGuard! Violations: {e.violations}")```+`threshold` is a violation probability in [0, 1], and a value outside that range raises `ValueError`. For a `score` question such as `harm_severity`, it applies to the probability that the level is at or above the middle of the scale (`serious` or `severe`), not to the expected level in `score`, so a mostly `minor` answer does not block on its own.+---## 3. Calibrated Confidence Gating