Jev directory
CommunityJev Benchmarks & Evaluations
Open-Jev Dataset
Hugging Face dataset of synthetic typed-decision records for Jev-style control and probability-estimation tasks. It holds 12 subsets with question, options, target, state_json, and metadata_json fields across train, calibration, validation, test, and ood splits.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- 12 subsets, including release-v2-redistributable at 114k rows and browser-drone-expansion-v1-redistributable at 161k rows.
- Splits include train (79.1k rows), calibration (4.67k), validation (3.72k), test (10.4k), and ood (15.7k).
- Each record carries id, group_id, split, source, kind, question, options, target, state_json, metadata_json, and record_json.
- Questions run 38 to 926 characters, with 2 to 16 options and 2 to 16 target entries per record.
- Licensed cc0-1.0 and stored as Parquet loadable through the Datasets library, pandas, or Polars.
Questions about this entry
Short answers drawn from the linked project or page.
It is a Hugging Face dataset with files auto-converted to Parquet and an API plus SQL console in the Dataset Viewer. The card lists Datasets, pandas, and Polars as supported libraries, so it can be pulled as a dataset object or read directly as Parquet.
The dataset card lists cc0-1.0 as the license. It is published as a public Hugging Face dataset, so access is through the Hub rather than a paid product.
The card lists English, Chinese, and Turkish. The sample rows shown in the digest are English-language customer support conversations, and per-record metadata includes a language field such as en.
No. Rows are synthetic. Metadata and record_json fields carry control_latents values such as ambiguity, churn, frustration, refund, repro, and severity, with provenance annotation status marked synthetic in the card sample.
Five splits are listed: train at 79.1k rows, calibration at 4.67k, validation at 3.72k, test at 10.4k, and ood at 15.7k. The calibration and ood splits support calibration and out-of-distribution measurement rather than training alone.
Records use a kind field, with choice and score among the values shown. A choice row carries an options list and a target such as [1,0,0,0], while a score row carries graded options with a target such as [1,0,0]. Question text is always included.
The dataset is hosted under the Hugging Face user allen65 and is marked as duplicated from ZefanCai/Open-Jev. Anyone depending on upstream updates should track the original repository, since this listing is a redistributed copy.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
Benchmark comparison of an untrained Qwen3.5-9B against hosted Jev across decision, tool-calling and calibration benchmarks. Reports accuracy, latency and ECE on matched tasks, plus a local adapter that serves Qwen through Jev's typed systemone API.
Jev Benchmarks & Evaluations#evaluation#calibration#benchmark
Spring Boot and Spring AI benchmark comparing Java keyword rules, Jev 1.13.0, and Claude Haiku 5.5 on warehouse goods-receipt triage, with a 200-remark synthetic dataset and a Python analysis script.
Jev Benchmarks & Evaluations#benchmark#evaluation#calibration
Communitymohammedahmadi1990/jev-triage
arXiv study applying Jev and Clef-Flash to manufacturing defect detection with as few as two reference images per class, including agentic refinement of text descriptions and probability calibration. Ships replication code, an R decisions package, and a no-code web app.
Jev Benchmarks & EvaluationsArticle#arxiv#calibration#classification
Community
Retrospective study comparing JEV, a non-generative decision model returning native option probabilities, with schema-constrained Qwen Flash on 2,048 patient-condition pairs from TrialGPT and MIMIC-III/IV under Zero Data Retention policies.
Jev Benchmarks & EvaluationsArticle#arxiv#evaluation#benchmark
Community
Hugging Face dataset of run results for the kieve-jev-fractal-300m decision engine, recording per-benchmark tallies of requests, answered, abstained and errored cases along with scores, latency medians and p95s, and retrieval and cost-aware metrics.
Jev Benchmarks & Evaluations#benchmark#evaluation#calibration
Community
An evaluation results dataset publishing per-benchmark outcomes for scairesearch/kieve-jev-moe-300m, including request, answered, abstained, and error counts, scored metrics, successful-request latency medians and p95s, and the runner configuration used for the run.
Jev Benchmarks & Evaluations#benchmark#evaluation#calibration
Community
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.