Back to benchmarks & research
metask-jev project preview
Benchmarks & research

metask-jev

Metask-Jev: calibrated typed-decision models (Jev-class). Single forward pass, candidate-logit readout. metask-jev-4b beats Bespoke Nimble-9B and Jev on JevBench.

Explore repository

How it uses Jev

探索最优开放权重的 typed-decision（Jev-class）模型：给定 state + bounded rubric，单次前向输出每个选项的校准概率。无生成、无解析。

Languages
Python 80% · Shell 20%
Last activity
Sep 2026
Added
2026-09-22
Stars at snapshot
2
Status
Community listing

Source: community catalog. Details and metrics reflect the published community snapshot and are not a performance endorsement. Documentation details extracted from GitHub · the README.

Listed in Jev Library

If this is your project, the badge below links back to this page. Paste it near the top of your README.

[![Listed in Jev Library](https://jevlibrary.dev/api/badge/metask-ai-metask-jev)](https://jevlibrary.dev/projects/metask-ai-metask-jev)

Something wrong on this page, or would you rather not be listed? Tell us and it will be changed.

  • jev-eval-agent project preview
    Benchmarks & research

    jev-eval-agent

    Personal-assistant agent built on Vercel's eve with 100 mocked tools, measuring how many steps it takes when Jev picks the tool versus the LLM.

    HTML 105Details
  • typesafe-ai-benchmark project preview
    Benchmarks & research

    typesafe-ai-benchmark

    This is a LLM Gateway that mimics typesafe ai structured output. Like an imposter Jev.

    TypeScript 38Details
  • jev-on-a-laptop project preview
    Benchmarks & research

    jev-on-a-laptop

    Unofficial study: Jev-style parallel typed decisions on stock 1.5B-8B models on an Apple Silicon laptop. Benchmarks, research notes, and a Hugging Face Space demo.

    Python 23Details
  • jev-benchmarks project preview
    Benchmarks & research

    jev-benchmarks

    Probability-aware evaluation for typed decision models: calibration, selective risk, latency, and reproducible benchmarks.

    Python 17Details
  • jev-lm project preview
    Benchmarks & research

    jev-lm

    A word-level language model whose output layer is Jev: n-gram drafter, Noul chunk verification, bits-per-token eval

    TypeScript 6Details
  • decider project preview
    Benchmarks & research

    decider

    One-pass typed decisions with calibrated probabilities (System One style model), fine-tuned from Qwen3.5-2B

    Python 351Details