Automated README summary
Automatically summarized from the README on 14 Oct 2026; not a run by jevbooks. Run instructions and author-reported outcomes still need verification.
What it is
Bespoke Nimble is an open 9B model (LoRA on Qwen3.5-9B) that takes text plus a flat schema of enum or boolean questions and returns one allowed answer token per field with probabilities. It ships with its data pipeline, training config, and eval harness under Apache-2.0.
How it uses Jev
The README does not describe using Jev in code. It only states Nimble is inspired by Jev's System One approach and reports 90.1% on its 324-example holdout against Jev's 93.2%. No Jev API calls or primitives appear in the README.
Technique worth stealing
Score one allowed answer token per field by reading logits for each choice code and softmaxing them, avoiding generated JSON parsing.
Try it
git clone https://github.com/bespokelabsai/nimble.git nimble; cd nimble; use Python 3.12.
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-1.13.0- 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 material0.74
- Shows a System One patternunclear0.55
- Describes handling uncertaintynot found in the excerpt0.05
- Reports measured numbersFound in material0.94
- Is a runnable toolnot found in the excerpt0.38
- Top-10 learning shortlist signalunclear0.41
- Is a replica of the modelFound in material0.96
- Solves a common problem (0–2)score on a 0–2 scale1.23 / 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 14 Oct 2026.