Automated README summary
Automatically summarized from the README on 9 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
CARVE is an Apache-2.0 Jev-like model family built on Qwen3.8-27B. It scores candidate letters for user-defined options and returns typed Noul, Choice, or Score answers through a TypeSafe-compatible System One API for decision tasks.
How it uses Jev
For each typed question, CARVE maps options to candidate letters, prompts Qwen3.8 with thinking disabled, reads vLLM candidate-token logits, and applies a restricted softmax over only those letters. The resulting Noul, Choice, or Score response is returned to a TypeSafe-compatible client.
Primitives:choicescorenoul
Technique worth stealing
Restrict the softmax to candidate option letters so each decision costs exactly one output token, avoiding free-form generation and JSON parsing.
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 material0.94
- Shows a System One patternnot found in the excerpt0.18
- Describes handling uncertaintynot found in the excerpt0.05
- Reports measured numbersnot found in the excerpt0.10
- Is a runnable toolFound in material0.67
- Top-10 learning shortlist signalnot found in the excerpt0.32
- Is a replica of the modelFound in material0.94
- Solves a common problem (0–2)score on a 0–2 scale1.01 / 2
- About JevFound in material0.93
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 9 Oct 2026.