Automated README summary
Automatically summarized from the README on 8 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
S1Q quantizes open Jev-like System One decision models that map a state and typed questions directly to finite choice, Boolean or discrete-level distributions. It supports pinned Kev, NanoJev, Laya, Intern-Decision and StartLux-Decision adapters, preserving native candidate order and decision readouts.
How it uses Jev
S1Q quantizes models that map a state and typed questions to finite choice, Boolean or discrete-level distributions. It uses teacher decision top-one versus runner-up logit margins to sample calibration rows, searches channel scales and clipping using actual quantized reconstruction, and fits a bounded ridge correction for activation error. The result is a quantized model that preserves native candidate order and decision readouts.
Primitives:choice
Technique worth stealing
Margin-Jacobian token proxy for calibration row selection, combined with activation-error compensation and joint weight/activation reconstruction.
Try it
Install with `python -m pip install -e '.[plots]'`, then run `python scripts/build_component_analysis.py --source-root results/benchmarks/expanded-20261007 --output-dir work/component-figures`.
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-1.13.0- Project signals
jev-1.13.0- Category
jev-1.13.0- Pattern tags
jev-1.13.0
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 material0.67
- Shows a System One patternnot found in the excerpt0.14
- Describes handling uncertaintynot found in the excerpt0.06
- Reports measured numbersnot found in the excerpt0.15
- Is a runnable toolnot found in the excerpt0.11
- Top-10 learning shortlist signalnot found in the excerpt0.22
- Is a replica of the modelFound in material0.84
- Solves a common problem (0–2)score on a 0–2 scale1.00 / 2
- About JevFound in material0.83
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 8 Oct 2026.