0
用 Jev（TypeSafe System One 决策模型）把上传的学习材料变成可自动判分的自助考试：逐得分点判定、置信门控、错题本与薄弱点。支持 DeepSeek / 智谱 GLM / 通义千问 / Kimi / OpenAI / Claude 等自带 Key。Self-exam generator and point-level grader built on a decision model.
Evaluation & benchmarks — page 10
Evaluation harnesses, benchmarks and graders. Jev scores or chooses between candidate outputs with an explicit rubric.
320 active projects match this task, ordered by traction.
975 PLC tag Expressions in Under a Quarter of a Second
Evaluate Jev (TypeSafe System One decision model) on GAOKAO-Bench objective questions - accuracy, confidence calibration and latency over 1497 Chinese college-entrance-exam multiple-choice questions
Can a System One model (TypeSafe Jev, open-source Kev) replace an LLM-as-judge in agent guardrail callbacks?
Review public code and documents with Jev in Codex and Claude Code: ranking, batch screening, claim checks, spend limits, and evaluation tools.
Probabilistic soft-constraint evaluation for volunteer scheduling using TypeSafe Jev, with deterministic application policy.
Multi-agent customer support API where the LLMs write the text and a typed decision model makes the decisions. Jev (TypeSafe System One) handles routing, response evaluation and retry/escalation; FastAPI + Ollama. Includes a 60-ticket benchmark against LLM free-text and structured-JSON baselines.
Behavioural anti-cheat research bench for Minecraft (Paper): mining telemetry, typed LLM questions, and an offline benchmark against classic X-Ray heuristics. Shadow mode only, never bans.
Let Jev judge. Let your code decide. Python issue-routing sandbox with labeled replay, an offline demo, and Kev support.
Head-to-head evaluation of Laya (open weights) and TypeSafe Jev on email intent classification
Benchmark de triagem de revisões sistemáticas: JEV (TypeSafe) × DeepSeek V4 Flash × Qwen3-Reranker, contra o julgamento documentado de autores reais (SYNERGY+ v3.0)
Judgment layer for Stratify: asks TypeSafe's Jev model what static analysis could not prove, and moves each finding along the confidence ladder the engine already uses. Rust.
Benchmark for typed System One decision models (Jev, Laya, and whatever comes next): calibration against a noise floor, framing sensitivity, selective prediction, ordinal fidelity, interference, robustness. pip install sys1bench.
No project description available.
OpenAI-compatible judge endpoint for agent approval gates: typed judgements (Jev), any OpenAI-compatible model, or a rule file — fail-closed, calibrated, with a replay battery
Jev-powered decision workflows for Codex image generation. Explicit constraints, bounded repairs, and honest evaluation.
Open-source Chrome MV3 extension: user-defined rubric categories, Jev noul scores each feed post, hide matches. BYOK Typesafe.
The most-discussed Hacker News threads, each read comment by comment by Jev and reduced to one verdict. Live from Neon Lakebase Postgres with BM25 search.
Independent 100-question English benchmark of TypeSafe's Jev decision model (jev-1.13.0)
A runnable decision-first agent evaluation example using Jev and TypeSafe AI.
No project description available.
Early evaluation of TypeSafe's Jev on clinic call triage: 98.6% accurate when confident, 116k decisions per dollar
Jev CLI for the jev-judge Claude/Codex skill. Wraps the official @typesafe-ai/sdk.
Prioritize human labels with active-learning strategies, durable review history and clean holdout evaluation.