AgenticRAGの判断をJevにしたら、爆速にはならなかったけど主導権が返ってきた
@qiita-trending.bot.chrs.to
@kevindubois.comI wrote a follow up on my #Jev + #LangChain4j blog post since I was able to sign up and get an actual API key this time 😄. (and added a few more features to the project)
@isev.org🆕New in #JEV🆕 Shear Stress-Stimulated Trafficking of Endothelial Piezo1 to Splenic Capsule Boosts Spleen Contraction and Exercise Endurance #extracellularvesicles #exercise #piezo1 #shearstress isevjournals.onlinelibrary.wiley.com/doi/epdf/10....
@sabyabhoi.bsky.social> General LLMs: can do anything, including structured output if the prompt holds up
> #Jev: can only output booleans, enums, and numbers
Guess which one is cheap enough to run on everything
@kuro-solutions.bsky.socialJEV:
→ Understand the query
→ Represent meaning beyond keywords
→ Retrieve and connect relevant knowledge
→ Let the model reason over a better knowledge space
Think of it this way:
Generative AI = reasoning + generation
JEV = knowledge representation + retrieval intelligence
#jev
@thedailytechfeed.comOpenAI’s Decisions API targets Jev-style agent monitoring, cutting cost and boosting speed. #AI #AgentMonitoring #DecisionsAPI #Jev #OpenAI #Security thedailytechfeed.com/openais-new-...
@global-news-2024.bsky.social📰 OpenAI’s Jev clone could help the frontier lab stop its swarming agents
OpenAI's "Decisions API" is a Jev clone that confirms the importance of fast, cheap intelligence.
Source: TechCrunch | Read story
💬 What's your take on this tech shift?
@michabbb.bsky.social
@andreygubanov.bsky.socialCan we use a System One model in the browser? I tested Laya, an open Jev alternative:
• 330–480 MB first load, then 2–5 s
• ~0.4 s per call on CPU, 0.13–0.21 s on GPU
• 7–14× slower in background Chrome tabs
@moellus.wandzeitung.xyz.ap.brid.gy#yolops von #jev bekräftigt, interessanter Ansatz https://www.pulumi.com/blog/self-driving-infrastructure-pulumi-jev/ /via #pulumi
@jihnma.bsky.socialTried a small Jev vs LLM experiment for adaptive interviews.
Same 100 synthetic interviews, same inputs, different judge.
Jev: 4.9s median / AUC 0.993
Claude Haiku 4.5: 40.4s / AUC 0.969
results:
github.com/jihnma/lab-j...
@sabyabhoi.bsky.socialEver since I found out about #Jev, half the problems I think about have turned into "is this a yes or a no"
@waynerad.bsky.socialJev vs Laya. (And some other models.) "Decision models under pressure."
"Seven systems do the same job: take a piece of text, a question, and a list of candidate answers, and return a probability over those candidates."
@blocksai.bsky.socialWhen TypeSafe released #Jev, it felt familiar. We toyed with a similar classifier design years ago. Our 'replica' implements Jev's API with a base model & runs within TypeSafe's published latency range, improving latency & accuracy for agents - and playing Doom.
Here's how: https://bit.ly/4iJsn4K
@galloni.netEveryone is suddenly talking about JEV — but what is it actually useful for? I tested the idea by building something real for less than $1. Here’s what happened. https://netcontentseo.com/article/what-is-jev-why-is-everyone-crazy-about-it-what-i-built-for-less-than-1-1059 #JEV #AI #AIAgents (1/2)
@ksarrazin.bsky.socialAwesome #Jev, une liste organisée de démos, outils et intégrations pour #TypeSafe Jev github.com/Amal-David/a...
@isev.org🆕New in #JEV🆕 Extracellular Vesicle-Encapsulated Small RNAs From Clostridium Butyricum Modulate Host Intestinal Lipid Metabolism #extracellularvesicles #gutmicrobiota #lipidmetabolism #obesity isevjournals.onlinelibrary.wiley.com/doi/epdf/10....
@daspoltergeist.bsky.social