nokia-applied-research/AnyJev
Turn any LLM into a Jev-style decision model: typed decisions, real probabilities, no training. (continue updating)
About the project
Turns any LLM into a decision model with typed questions (choice, yes/no, score) that return real probabilities read from the next-token distribution, without training. L0 removes position and label-prior bias; L1 adds temperature calibration from 100-500 labels.
Useful for
- Route incoming requests to a handler with a thresholdable confidence instead of raw logits.
- Gate a proposed tool call by asking whether it is destructive or irreversible.
- Calibrate a yes/no question on 100-500 labeled examples to get a usable p_true.
README summarized by DeepSeek V4.1 Flash. Details may be inaccurate.
Why it’s trending
- 99 stars so far today, about 205 expected by the end of the day.
- The spike has held for 3 days in a row — not a one-off blip.
- The repository is 2 days old and already has 271 stars. With less than two weeks of history, there's no usual pace to compare the spike against yet.
- Top new repositories this week: #77.
- About 66 forks a day — people are taking the code.
Stars per day
Bars are daily stars, the line is the usual pace. Red marks spike days.
Numbers
- Total stars
- 271
- Today
- 99 · ≈ 204 by evening
- Forks
- 39
- Issues and pull requests
- 1
- Watchers
- 6
- Language
- Python
- License
- Apache-2.0
- Latest release
- v0.0.2 · September 21, 2026
- Created
- September 21, 2026
- Last push
- September 23, 2026
Star trust
Growth looks organic: forks and discussion are in line with active projects, and stars arrive unevenly, the way people give them.
These are heuristics, not a verdict: we judge by the repository’s behavior, not by a list of stargazers.
Spotted in
- September 23, 2026Top new repositories this week: #77
Share
README badge
Paste it into your README — the badge shows the current magnitude and links to this page.
[](https://gitnova.dev/en/r/nokia-applied-research/AnyJev)Similar by description
-
8.8
NandhaKishorM/laya
Laya is a non-autoregressive decision engine that answers typed questions (choice, score, noul) over any state (text, email, ticket, JSON) in a single forward pass (~35 ms on GPU) with calibrated probabilities.
-
2.1
kshetrajna12/reflex
An open decision model: given a state (text, JSON, or image) and typed questions, it returns calibrated probabilities over fixed answer options. Runs on top of Qwen3.5 and answers with numbers instead of free text.
-
2.6
featherless-ai/simple-jev
A Python server that turns open Hugging Face LLMs into a classifier: it reads next-token logits and returns JSON with choices, rubric scores, or truth judgments without training a separate classifier head.
-
1.0
y0usaf/pi-jev
An extension for the Pi coding agent that adds a TypeSafe Jev decision layer: it judges bash/write/edit calls before they run and inspects command output, returning typed probabilities instead of prose.
-
6.1
malevrigns/agent-jev
AgentJev-0.6B is a specialist decision model for AI agent loops: it takes unstructured state (diffs, traces, logs) plus typed questions and returns a probability distribution over options in a single forward pass with…
-
3.4
razorback16/openjev
An open-source "System One" decision server on DiffusionGemma: send a state and typed questions (yes/no, choice, score) and get probabilities plus confidence in tens of milliseconds, with no text generated.…