jevbooks

← All projects

simple-jev

featherless-ai/simple-jev

Turn open models into typed classifiers without a trained head.

Classification & Routing79%Runner-up: Evaluation & BenchmarkingMany questions, one request

What it is

Simple Jev lets you use compatible open models from Hugging Face for structured classification and scoring, without training a separate classifier head. It is for developers who want typed decisions from a model's next-token logits.

How it uses Jev

The server reads the model's next-token logits for each question and builds a JSON response with choices, rubric scores, or truth/support judgments. The model does not generate a JSON completion; the server constructs the response from the scores. Shared request validation, versioned prompt instructions, and response scoring live in the plain Python common/ folder.

Primitives:choicescorenoul

Technique worth stealing

Read next-token logits for each question and construct typed JSON responses without generation.

Try it

Use the public demo API with no login, or run the HF server locally with Python 3.12+.

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-centricyes0.93
  • Shows a System One patternyes0.83
  • Handles uncertaintyno0.05
  • Measuredno0.05
  • Runnableyes0.91
  • Worth recommendingunclear0.40
  • Model replicayes0.66
  • Problem scopescore on a 0–2 scale1.08
  • About Jevyes0.92

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