User repository on GitHub
Using Apple's Neural Engine as a fast and free local "System One" Decision Engine (macOS 27) - "Honey, we have Jev at home"
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Tools & CLIs
systemANE
Two-tier decision engine pairing a 22.6M-parameter MiniLM encoder on the Apple Neural Engine with the on-device Apple Foundation Model as fallback for uncertain inputs.
- Stars
- 15
- Language
- Python
- Published by
- Community
- Author
- kerryrm
- Last verified
- Oct 2026
About this project
Typed decisions on the Apple Neural Engine. 1.4 ms each, with 94.6% of the graph executing on the ANE, from a 22.6M-parameter model with no fine-tuning. On CLINC150 — 1,300 human-written utterances collected independently of this project, 300 in-scope plus 1,000 genuinely out-of-scope — that gets 93.3% accuracy, ECE 0.041, and 0.994 AUROC at spotting out-of-scope input. It is a System-1 decision engine: typed decisions over a sentence encoder, paired with a second tier that only runs when the first is not sure.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- Tier 1 answers in about 1.4 ms with most of its graph on the Neural Engine per the planner script.
- Second tier runs through the OS-bundled Foundation Model CLI over plain HTTP with nothing external.
- One encoded state serves several typed questions, so three decisions cost little more than one.
- Ships scripts measuring device placement, fp16 drift, warmup penalties, and serving throughput.
Watch out
Tier 2 needs the OS-bundled Foundation Model CLI, and cold-start gaps raise per-call latency above the saturated figure. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
No. The accuracy lives in the second tier; tier 1 is deliberately simple and its job is to answer cheap inputs fast and refuse or escalate the rest.
Mostly yes for this workload: end-to-end fp16 versus fp32 comparison broke no correct in-scope answers on one suite and one token on the other, with the loss already uncertain in fp32.
Expect around 3 ms instead of 1.4 ms when calls arrive every few hundred milliseconds, since the encoder goes cold between calls; the ANE lead over CPU and GPU widens there.
Out-of-scope inputs are refused rather than escalated, because the second tier is bound to the same labels and would be forced to pick a wrong one.
With Python 3.12, install requirements, run build.py once to convert MiniLM into encoder.mlpackage, run where.py to confirm it lands on the ANE, then start fm serve and run cascade.py.
The Apple Foundation Model, roughly 3B parameters, served by the fm serve CLI built into macOS 27 at /usr/bin/fm, which fmserve.py talks to over stdlib HTTP only.
serve.py reports 708 requests per second at 16 clients on a laptop with p50 1.21 ms, against decider-2b's published 431 req/s at 64 clients on an NVIDIA GH200, though the tasks differ.
More Jev resources like this
34GitHub stars
Pure-HTTP automation for the TypeSafe register, magic-link login, onboarding, and API-key creation flow, with pluggable temp-mail backends, concurrent batch runs, and key verification exports.
Jev Tools & CLIs#community#awesomejev#catalog-import
Community
2951461586/Jev-Register-ToolPython
2951461586/Jev-Register-ToolPython36GitHub stars
Python toolkit plus 0.8B multimodal model that takes a message, document, or image and returns calibrated choice, yes/no, and score answers in one forward pass for local decision workloads.
Jev Tools & CLIs#community#awesomejev#catalog-import
Community
PsiACE/dohnutsPython
PsiACE/dohnutsPython39GitHub stars
Video tool where Jev scores subtitle segments and code assembles a short highlights cut plus a de-watered full cut with timestamped summaries.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
cclank/jevclipPython
cclank/jevclipPython30GitHub stars
Forkable weekday radar that asks Jev one yes-or-no question per interest over each new arXiv announcement and publishes a page plus RSS feed.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
Eliot5566/JEV-Paper-RadarPython
Eliot5566/JEV-Paper-RadarPython38GitHub stars
Async supervisory layer for Hermes Agent that locks Definitions of Done, proves mechanical criteria locally, and calls Jev only for unresolved semantic checks under a token budget.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
keeltrace/hermes-nervePython
keeltrace/hermes-nervePython19GitHub stars
Windows desktop aid that reads the local WeChat database for the foreground chat, strips identifiers, and asks Jev for stage, intent, tension, suggested moves, and a reply review.
Jev Tools & CLIs#community#jevbest#catalog-import
Community
yushen100/wechat-jev-assistantPython
yushen100/wechat-jev-assistantPythonFrom the Jev guides
Original write-ups on what this entry is for.
The request shape barely changes across surfaces; the surrounding contract does. When to call the API directly and when a gateway, a Worker binding, or a router entry earns its extra hop.
Jev Tools & CLIs#gateway#vercel#cloudflare
Read article
Taking a decision checkpoint out of Python: what the GGUF and browser WebGPU exports preserve, what they change, and the parity evidence the model cards publish.
Jev Tools & CLIs#open-models#local#onnx
Read article
Jev labels a record that already exists. Code moves, flags, or holds it. How the document, mail, and transfer tools keep the action out of the model.
Jev Tools & CLIs#classification#email#security
Read article
From the Jev community
Posts from builders shipping with Jev right now.