Skip to content
All builds /Research and data/openjev: Local Non-Autoregressive Decision Engine
github·Research and data

openjev: Local Non-Autoregressive Decision Engine

An open-source PyTorch / vLLM implementation experimenting with RLCD (Reinforcement Learning for Calibrated Decisions) on small base weights.

Open GitHub (openjev) ⚡ 1800 rps · 3.2 GB VRAM
openjev: Local Non-Autoregressive Decision Engine
OpenJev Community

OpenJev Community

@openjev

OpenJev v0.2: Replicating System 1 parallel heads on edge hardware. 1,800 decisions per second on consumer GPUs.

01 / The Decision

Classify document sentiment and topic category simultaneously.

Jev sits between the incoming context and the next system action. Rather than generating lengthy, slow natural language that requires brittle regex parsing, Jev returns non-autoregressive, calibrated probabilities that downstream code can immediately execute.

02 / Typed Outputs
01sentiment_probPrimary action
02topic_vectorConfidence score
03uncertaintyFallback route

Related Research and data Builds

xpost

Search and tagging on keep.md

Cloudflare Workers has Jev now so I'm putting it to the test on keep.md - 7x faster search rerank compared to the current hybrid - 50x faster tagging of content vs GLM 4.7 Flash with no failures

Inspect
xpost

A visual reference finder

I built a visual reference finder with Jev One single prompt → 100 images from Cosmos, NASA, and The Met my new rabbit hole for creative work 🌻

Inspect
xpost

Classify rows in DuckDB

I made a DuckDB extension where you can use @typesafeai 's Jev to do quick classification of rows in any csv/parquet file or duckdb table about 10sec for 1k rows ~ better than using an LLM, way more ergonomic than a classifier game-changing for data analysis!

Inspect