All projects

laya-guardrails

javimp2003/laya-guardrails

Guardrails de input, tool call y output para agentes de IA con un modelo System One tipo Jev (laya-pt-es-typed) autoalojado en una NVIDIA L4: 5 ms por check frente a 140 ms de un LLM-as-a-judge.

Python
Stars
0
Forks
0

Why it is listed

Self-hosted agent guardrails use a Laya decision model for input, tool-call and output checks, following a Jev-like typed System One approach rather than hosted Jev.

Related projects

  • Decis

    Self-hosted, Jev-compatible decision-model API — one /v1/systemone endpoint, open weights (Laya, kev), one Docker image per engine.

    Shared topics: fastapi, laya

  • OpenJevSwift

    Native Swift implementation of OpenJev, the Jev-compatible typed-decision server: yes/no, choice and score answers read from model probabilities, on Apple silicon Macs and in iPhone apps. Serves Verdict, Laya, DiffusionGemma and JevK5. Apache-2.0.

    Shared topics: laya

  • usejev

    Run Laya locally with Bun: native ONNX inference, a TypeSafe-compatible API, and a bilingual decision playground.

    Shared topics: laya