Automated README summary
Automatically summarized from the README on 25 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
jev-tune is a calibrated decision and safety shell demonstrated on PID autotuning. It answers typed questions about gain steps, response quality and step health, then applies temperature scaling, conformal label sets and rollback. It targets engineers tuning PID loops, including in MATLAB/Simulink models.
How it uses Jev
The head answers three typed questions on the measured closed-loop response state: choice (which gain step), score (how good the response is) and noul (whether the next step is healthy). The shell temperature-scales these answers, builds split-conformal label sets, can hold an axis before actuation when its set is empty, too wide or contradictory, and a deterministic SafetyGate rolls back after a bad measurement.
Primitives:choicescorenoul
Technique worth stealing
Treat the decision as calibrated probabilities: temperature-scale typed answers, wrap them in split-conformal sets, and abstain or roll back when the set is unusable.
Try it
git clone the repo, pip install -e ., then run jev-tune demo --plants 12.
Model judgments from README excerpts
Project signals use repository metadata, an existing summary and the first 700 README characters, not the full project. “Not found” means not detected in that material; it does not establish absence. These are the original judgments with recorded models and dates, not source audits or run results.
Models & evaluation dates
- Inclusion gate
jev-1.14.0- Project signals
jev-1.14.0- Category
jev-1.14.0- Pattern tags
jev-1.14.0
Evidence
These signals use repository metadata, an existing summary and at most 700 cleaned README characters. ≥ 0.60 means reported in that excerpt; ≤ 0.40 means not found there. The middle band is inconclusive. None is a source-code audit or a test result. The runnable-tool question excludes research experiments; the recommendation question asks for a learning shortlist, not a quality verdict.
- Jev is core to the projectFound in material0.96
- Shows a System One patternFound in material0.93
- Describes handling uncertaintyFound in material0.86
- Reports measured numbersnot found in the excerpt0.07
- Is a runnable toolFound in material0.66
- Top-10 learning shortlist signalFound in material0.61
- Is a replica of the modelFound in material0.72
- Solves a common problem (0–2)score on a 0–2 scale1.11 / 2
- About JevFound in material0.84
no-judgment band (0.40–0.60)
no-judgment band (0.40–0.60)
The Jev model and date for each judgment stage are listed above. Card written by DeepSeek V4.1 Flash from the README on 25 Oct 2026.