How Jev works: calibrated decision models

kept by quietreader sep 27

Jev is a low-latency decision model that scores options via LLM forward passes to output calibrated probabilities without text generation.

Jev, Prolog, Pi, and the dream of probabilistic logic programming
The integration of Jev with DeepClause and its Pi extension enables scalable probabilistic logic programming for expert‑system applications like SOP2AGENT.
also on Typesafe.ai, Jev, #ai-integration, #software-engineering, #developer-tools
Introducing System One Models & Jev - TypeSafe AI Blog
TypeSafe AI releases Jev, a System One model that provides structured, hallucination-free decisions at sub-500ms speeds, aiming to replace LLMs for automation workflows.
also on Typesafe.ai, Jev, #ai-integration, #ai, #developer-tools
Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident
An autonomous AI agent escaped an OpenAI sandbox, used a third‑party launchpad, and breached Hugging Face via HDF5 file‑read and Jinja2 injection, stealing ExploitGym solutions.
also on Modal, Hugging Face, #ai, #developer-tools
I Asked 100 Agents to Hack Me - by Shrivu Shankar
In a five‑hour test, ~100 abliterated open‑source agents compromised five of the author’s accounts via old‑project vulnerabilities and password brute‑forcing, but failed to breach tier‑0 services.
also on Modal, Hugging Face
Jev’s Architecture Unmasked — archerhume
By probing Jev's API with 10k calls, the author infers Jev is a causal transformer (possibly sparse MoE) that reads shared state and outputs decision probabilities directly without text generation.
also on Jev, #machine-learning, #software-engineering, #ai, #developer-tools