jevbooks

← All projects

minojev

zeredy879/minojev · Homepage

Typed calibrated decisions from hidden states, zero output tokens.

What it is

minojev is an open replica of a decision model that answers runtime-defined Choice, Boolean, and Score questions with calibrated probability distributions in one forward pass, for developers needing fast, parse-free decisions.

How it uses Jev

Jev makes typed decisions (choice, boolean, score) from a state and questions, returning calibrated distributions. The result is used directly in code without token generation or parsing.

Primitives:choicescore

Technique worth stealing

Freeze backbone, cache candidate features, train only a small decision head with distribution losses.

Try it

uv venv --python 3.12 .venv && . .venv/bin/activate; uv pip install -e ".[test,monitor]"

View on GitHub

judged by Jevjev-1.13.0

Evidence

Each line is one question put to Jev about the README. ≥ 0.60 reads as yes, ≤ 0.40 as no; in between Jev is not making a call.

  • Jev-centricyes0.72
  • Shows a System One patternno0.26
  • Handles uncertaintyno0.04
  • Measuredyes0.67
  • Runnableno0.35
  • Worth recommendingno0.30
  • Model replicayes0.90
  • Problem scopescore on a 0–2 scale1.31
  • About Jevyes0.63

Signals by Jev jev-1.13.0, card written by DeepSeek V4.1 Flash from the README on 20 Sept 2026.