7n1m4/jevArchived
Organization repository on GitHub
Local CPU-only TypeSafe System One decision service + oh-my-pi plugin
- Forks
- 0
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Integrations & Adapters
Jev Local — 7n1m4
A local, CPU-only Jev-Style-0.8B-Decision-v3 service with an oh-my-pi extension that points the agent's typesafe judge provider at it over loopback HTTP. No GPU, no API key, and the 505 MiB model is fetched separately.
- Stars
- 0
- Language
- Shell
- Reading time
- 2 min
- Published by
- Community
- Author
- 7n1m4
- Last verified
- Oct 2026
This is one of several Jev listings for the same project. The canonical entry is Jev Local — Argos1111.
Used inJev for Software & DevTools
Highlights
- Serves the oh-my-pi systemone / alpha/decisions contract over loopback HTTP, so judgments never leave the machine.
- The scorer reads llama.cpp logits directly as logit[' yes'] minus logit[' no'] at each ' ->' slot; no text is generated.
- The extension registers the provider with an explicit baseUrl, the mechanism that retargets the agent's judge role.
- systemd/jev-style-decision.service runs as a --user unit, so it survives logout and comes back after a reboot.
- The 505 MiB model is not in the repo; the README has fetch and uninstall steps, and the committed unit omits MemoryMax.
Watch out
README reviewed 2026-10-17; the repository lists no license and the captured README cuts off inside the readout spec, so the full readout_config.json format, the model fetch commands and any benchmark numbers are not available.
Questions about this entry
Short answers drawn from the linked project or page.
A local decision service for Jev-Style-0.8B-Decision-v3 plus an extension that points the oh-my-pi agent's typesafe judge provider at it. It runs CPU-only over loopback HTTP with no GPU, no cloud calls for judgments and no API key.
The scorer reads decisions straight out of llama.cpp as the logit for ' yes' minus the logit for ' no' at each ' ->' slot. No tokens are sampled, so there is no generated text to parse and no generation latency.
Jev-Style-0.8B-Decision-v3 is a decision model, not a chat model: it reads a state plus a typed question, noul, choice or score, and answers. Its readout spec lives in readout_config.json under the macjev-readout-v1 format.
The README's headline is CPU-only: no GPU, and judgments run locally, so no API key is required. The 505 MiB model file is not committed to the repository and must be fetched following the README.
systemd/jev-style-decision.service installs as a --user unit, so it survives logout and comes back after a reboot. The committed unit is portable (%h, no MemoryMax); the installer writes a MemoryMax measured on the target machine.
It retargets the agent's judge role at a local service instead of the cloud: the extension registers the provider with an explicit baseUrl. The README frames this as no cloud calls for judgments and no API key, not as a hosted replacement.
It is 7n1m4's repository on the main branch, created and last pushed 2026-10-13 with no stars, no forks and no license listed; the README documents a service, a scorer, an extension and systemd units, but no benchmarks.
More Jev resources like this
A frozen decision server that serves Gemma 4 31B — quantized to FP8 at startup — in TypeSafe's /v1/systemone format, reading the next-token distribution instead of generating tokens. It needs one FP8 NVIDIA GPU with 80 GB and has no authentication.
Jev Repos & SDKs#community#github#python
Community
krishna-gogineni-765/deck31bPython
krishna-gogineni-765/deck31bPythonThe nano tier of local Jev: a /v1/systemone-compatible typed-decision server — noul, choice and score questions with probabilities — running a 7M-parameter model CPU-only from an 8MB tarball. Point any Jev/TypeSafe SDK client at it via base_url.
Jev Models & Reimplementations#community#github#local
Community
javimosch/jev-nanoPython
javimosch/jev-nanoPythonHTTP server for the Mica v0.5 4B Preview decision model: GGUF weights run through llama.cpp and answer noul, choice and score questions with calibrated probabilities over the TypeSafe /v1/systemone format.
Jev Repos & SDKs#community#github#python
Community
akivet/MicaPython
akivet/MicaPythonServing code for the Jeff 1.0 family of Jev-style decision models: a state, a typed question and its options go in, and one forward pass returns a probability for each option. Four open models from 4B to 31B parameters are served the same way.
Jev Repos & SDKs#community#github#python
Community
j-geuter/jeff-servePython
j-geuter/jeff-servePythonA GitHub App that routes pull-request review work through Jev: it ingests the webhook, normalizes the PR context, gets a LOW, NORMAL or SPECIALIST decision, and a deterministic policy layer maps it to auto-approval, a review request or a check-run update.
Jev Integrations & Adapters#community#catalog-import#logicrw
Community
hexyedev/jevprPython
hexyedev/jevprPythonAn auto-mode plugin for the Pi coding agent that classifies shell commands before they run: built-in rules decide most calls, and Jev (hosted) or Kev and Laya (local, CPU, offline) route the rest to run, confirm, or block.
Jev Integrations & Adapters#community#github#pi
From the Jev guides
Original write-ups on what this entry is for.
From the Jev AI model community
Posts from builders shipping with Jev right now.