Jev in Search: Three Practical Evaluations
Independent experiments on Jev for search stopping, memory reranking and multi-hop relation selection, with stated limits.
This summary is based on how curated lists describe the project; its README was not available to review.
Overview
Jev in Search: Three Practical Evaluations is a blog post reporting independent experiments on three search-related decisions: when to stop searching, reranking agent memory, and selecting relations for multi-hop retrieval. It links to implementations.
The list summaries note limitations including private data, unequal sample counts and a simulated speed illustration. Specific results were not collected here.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Yes / no | Is there enough evidence to stop searching?(inferred) | — |
| Score | How relevant is this memory item?(inferred) | — |
| Choice | Which relation should the next hop follow?(inferred) | — |
How it works
- Set up three retrieval tasks
- Insert Jev at the decision point of each
- Compare with baselines
- Report results and limitations
Strengths
- Covers several distinct retrieval decisions
- Implementation links provided
Caveats
- Private data and unequal samples
- Speed illustration is simulated
At a glance
- Type
- Article
- Status
- research
- Has code
- yes
Use cases and integrations
How curated lists describe it
Independent experiments on search stopping, memory reranking, and multi-hop relation selection, with implementation links and limitations including private data, unequal sample counts, and a simulated speed illustration
AnotiaWang/awesome-decision-modelsIndependent experiments on search stopping, memory reranking, and multi-hop relation selection, with implementation links and limitations including private data, unequal sample counts, and a simulated speed illustration.
cobanov/awesome-jevIndependent experiments on search stopping, memory reranking, and multi-hop relation selection, with implementation links and limitations including private data, unequal sample counts, and a simulated speed illustration.
hellogumbo/awesome-jev
Frequently asked questions
What is evaluated?
Search stopping, memory reranking and multi-hop relation selection.
Are results reproducible?
Partly; some data is private, as the summary notes.
Is the speed comparison real?
The summary says the speed illustration is simulated.
Related projects
Local long-term memory for coding agents with an optional Jev reranker that orders recalled memories.
ProjectproductionTypeScriptVolcengine's context database for agents, with a Jev rerank client that scores candidate documents with jev-latest as relevance.
LibraryproductionPythonTerminal tool that finds link paths between Wikipedia articles: Jev ranks each page's outgoing links while Python runs the search.
ProjectbetaPythonFilters an assistant's memories by relevance, not resemblance: one Jev yes/no per memory in a single request, kept above a threshold.
ProjectbetaPythonMulti-user MCP 'agents' cloud' on Cloudflare Workers with an optional search stage that reranks candidates with Workers AI Jev scores.
MCP serverbetaTypeScriptMilvus-based multi-hop graph RAG library where two-stage Jev Noul scoring selects relations and reranks passages.
LibrarybetaPython