Reproducing Jev's ideas from scratch: a parallel decision system with shared KV caches on Qwen3 1.5B
Sam_Witteveen · x · 2026-09-20
A developer reproduced ideas behind Jev from scratch, starting with a tiny Qwen3 1.5B model and building a parallel decision system using shared KV caches. They benchmarked sequential vs batched inference and achieved 100% decision agreement in cache correctness tests, sharing lessons in a thread.
More from coding & agent
- mitsuhiko on the Shared Frustrations of Agentic Software Engineering — mitsuhiko · 2026-09-20
- Comprehensive 55-minute Codex Desktop Tutorial Covers Skills, MCP, TikTok and Blender — aziz4ai · 2026-09-20
- Stalkr adds keyword groups to benchmark your brand vs. competitors, with API and MCP access — marclou · 2026-09-20
- Proval: open-source self-hosted LLM code review agent in a single Docker container — Dazzling_Cancel4505 · 2026-09-20
- Jev agent demo: classifies bribes, threats and pleas with no keyword matching — chongdashu · 2026-09-20
- Parallel structured LLM answers never check each other: the Zhaozhou MU problem — Successful-Farm5339 · 2026-09-20