Automated README summary
Automatically summarized from the README on 13 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
A Python library, HTTP server, CLI, and training loop that answers typed questions (choice, score, noul) about text or JSON using a small NLI cross-encoder, for local, private, CPU-only decision making.
How it uses Jev
Jev makes decisions by treating entailment probability as option probability: choice uses max-pooled entailment per option then softmax; score uses entailment per level then expected value; noul uses entailment vs contradiction renormalized. Results are returned as calibrated probabilities via the TypeSafe POST /v1/systemone shapes.
Primitives:choicescorenoul
Technique worth stealing
Use a small NLI cross-encoder as a decision model: entailment probability is the option probability.
Try it
git clone && cd modernbert-jev; uv venv && uv pip install -e .; modernbert-jev pull modernbert-jev; modernbert-jev serve --model models/modernbert-jev --port 11436
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 material0.90
- Shows a System One patternnot found in the excerpt0.19
- Describes handling uncertaintynot found in the excerpt0.07
- Reports measured numbersnot found in the excerpt0.16
- Is a runnable toolFound in material0.91
- Top-10 learning shortlist signalunclear0.40
- Is a replica of the modelFound in material0.95
- Solves a common problem (0–2)score on a 0–2 scale0.94 / 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 13 Oct 2026.