Retrieval reranking and RAG with decision models

Experimental 17M-400M English decision models for choice, yes/no, score and reranking, with weights, training code and an ONNX browser demo.

Open Jev-style 0.6B decision model plugged into BlitzRank and Reranker-Guided Search, scoring k candidates per match in one pass.

DeepSeek Harness paper-reading workbench whose PDF search reranks keyword candidates with Jev, falling back to keyword order.

CLI and MCP server giving coding agents Jev-powered semantic search, focused reads and list picking in about a second for about a cent.

Web search CLI for agents that has Jev score results and scraped chunks against the query and goal, passing only relevant ones on.

Fourteen-dataset reranking study of Jev vs Cohere Rerank 4, zerank-2 and chat models, with raw responses and bootstrap intervals.

Graded relevance study over 9,831 labelled pairs: Jev reranking alone does not beat embeddings, but fused with them it is best.

Reproducible Turkish RAG benchmark comparing Jev batch-Noul reranking with Cohere Rerank on quality, latency and cost.

Reproducible Chinese RAG reranking benchmark for Jev on T2Reranking with hard negatives, staged and cost-capped, logs published.

Profile matching that encodes preference criteria as Clef or Jev decision vectors, retrieves with HNSW and checks rules both ways.

Auditable reranking benchmark: Jev vs Qwen3-Reranker-0.6B on BEIR SciFact and NFCorpus for nDCG, cost per query and tail latency.

Training-free local decision layer fusing four readings from two frozen Qwen models, auto-releasing only when readings agree.

Independent benchmark asking whether Jev can replace an LLM reranker, on natural-language search over 6,245 YC companies, with a live demo.

Milvus bootcamp notebooks combining Gemini embeddings, Milvus retrieval and Jev judgments for reranking, filtering, routing and stopping.

Repository and Rust crate search tool that asks Jev to score and select retrieved candidates for a request.

Test project using Jev to build chat-style search over a blog with hundreds of articles.

Official TypeSafe cookbook showing how to judge retrieved passages with Jev before passing them to an answering model.

Official TypeSafe cookbook recipes for guardrails, re-ranking, RAG passage classification, citation checks and more with Jev.

Sayora write-up comparing Jev with production embeddings for picking the right website page to answer caller questions.

TypeSafe cookbook showing how to evaluate many typed questions over one shared state in a single Jev request.

Independent experiments on Jev for search stopping, memory reranking and multi-hop relation selection, with stated limits.

Post reporting the treg team runs Jev in production for traffic classification, onboarding and reranked search.

Reddit thread asking whether Jev works as a retrieval relevance filter or reranker for RAG.

X post reporting a measured catalog-search run where Jev reranking alone did not beat vector retrieval.

Chinese X post measuring Jev in a RAG pipeline over TypeSafe's docs: correct top-ranked answers rose from 7 to 17 of 20.