Automated README summary
Automatically summarized from the README on 15 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
JevForge is an open training and inference stack for Jev-style decision models. It builds training records from states, questions, and candidate sets, trains and calibrates a scorer, evaluates fixed splits, and serves a Jev-compatible API for interactive web decisions.
How it uses Jev
JevForge expands one question into K rows, each with the same state and question plus one candidate path. It pools the final valid token, applies a two-layer GELU scorer to every candidate, then normalizes only the K logits for that question. The resulting distribution is served via a Jev-compatible API.
Primitives:choicescorenoul
Technique worth stealing
Normalizing only the K logits belonging to one question keeps candidate scores comparable and preserves uncertainty.
Try it
bash scripts/run_mac_demo.sh
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 material0.91
- Shows a System One patternunclear0.54
- Describes handling uncertaintynot found in the excerpt0.07
- Reports measured numbersnot found in the excerpt0.23
- Is a runnable toolFound in material0.70
- Top-10 learning shortlist signalunclear0.59
- Is a replica of the modelFound in material0.95
- Solves a common problem (0–2)score on a 0–2 scale1.00 / 2
- About JevFound in material0.97
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 15 Oct 2026.