jevbooks

← All projects

openjev

zhihz/openjev

Local bilingual probability decisions from context, questions, and candidates.

Scoring & Ranking50%Runner-up: Evaluation & BenchmarkingMany questions, one request

What it is

Open JEV is an independent research preview that answers bilingual probability questions from context, questions, and candidate answers. It is aimed at developers who need a local decision interface for applications.

How it uses Jev

Jev is not used. Open JEV is an independent project inspired by TypeSafe's Jev; it does not contain Jev's code, weights, or RLCD implementation. The current backend is a frozen Qwen3-4B-Instruct-2507 model that reads one next-token answer-label distribution per question.

Technique worth stealing

Reads a next-token answer-label distribution over supplied candidates instead of generating explanatory text.

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-centricunclear0.49
  • Shows a System One patternno0.15
  • Handles uncertaintyno0.04
  • Measuredno0.05
  • Runnableunclear0.56
  • Worth recommendingno0.12
  • Model replicayes0.78
  • Problem scopescore on a 0–2 scale0.76
  • About Jevyes0.76

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