Verdict-open-jev
Non-autoregressive decision engine on ModernBERT (151M) with calibrated uncertainty (RLCD), TypeSafe AI Jev benchmark audit, and in-browser WebGPU playground
How it uses Jev
OpenJev (Verdict) is an open-source, post-trained foundational decision model designed for structured software workflows, inspired by TypeSafe AI\'s Jev and Reinforcement Learning for Calibrated Decisions (RLCD) .
Use cases
- MCP Tool Selection & Subagent Skill Routing :
- Modern coding agents spend valuable context tokens and hundreds of milliseconds querying LLMs just to select an MCP tool (e.g. picking between git commit , search files , or read…
- Real-Time Agent Guardrails & Security Triage :
- Instead of spinning up a heavyweight LLM subagent to review each step for prompt injections or policy violations, OpenJev acts as a lightning-fast semantic gatekeeper at the perim…
- High-Frequency State Loops (Robotics & Interactive Simulation) :
- Evaluates game/sensor states (e.g. Doom, Starcraft, IoT sensor triggers) and selects discrete actions in real time for pennies per million executions.
Getting started
Install the library in your Python 3.10+ environment:
Source: community catalog. Details and metrics reflect the published community snapshot and are not a performance endorsement. Documentation details extracted from GitHub · the README.