User repository on GitHub
Per-row semantic judgements over large relational tables on a decision model (Jev)
- Forks
- 12
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Repos & SDKs
SOLO (System One Layout Optimizer)
A System One layout optimizer for decision-model workloads: it reorders rows and fields so a prefix-caching backend reuses input computation, then returns per-record decisions in original order. Its bundled backend serves AutoTrust JEV-9B through vLLM.
- Stars
- 792
- Language
- Python
- License
- MIT
- Reading time
- 2 min
- Published by
- Community
- Author
- 0814wdwd
- Last verified
- Oct 2026
Highlights
- Reorders rows and fields to expose shared input prefixes, then returns decisions in your original row order; each request still carries the complete record.
- Project benchmarks on one RTX 4090: up to 11.7x over original order with long shared context and 3.28x over NDV sorting with correlated fields.
- Serves AutoTrust JEV-9B, a Qwen3.5-9B-based open decision model, through vLLM with binary, categorical, or scored decisions and probabilities.
- Python 3.10+; install with pip install ".[pandas,hub]" from a source checkout and scan tables with DecisionEngine().scan().
- The layout optimizer also works with your own decision-model backend; the client downloads only the pinned model's two decision-head metadata files.
Quickstart
python
import pandas as pd
from solo_layout import DecisionEngine
tickets = pd.DataFrame([
{"id": "T-104", "policy": "Refunds within 30 days", "request": "Please refund my order."},
{"id": "T-201", "policy": "Refunds within 30 days", "request": "Can I exchange the size?"},
])
with DecisionEngine() as engine:
result = engine.scan(tickets, "Does the request ask for a monetary refund?")
print(result.to_pandas())Watch out
README reviewed 2026-10-17 and truncated in the deployment section, so GPU requirements beyond the single RTX 4090 test and the license terms of the JEV-9B weights are not in the digest. The 11.7x and 3.28x figures are the project's own reproducible synthetic workloads, not independent benchmarks.
Questions about this entry
Short answers drawn from the linked project or page.
The System One Layout Optimizer: an input-layout optimization tool that runs a natural-language decision over every record in a dataset, arranging rows and fields for prefix-cache reuse and returning decisions in your original order.
The included backend serves AutoTrust JEV-9B, an open decision model built on Qwen3.5-9B, through vLLM. It returns binary, categorical, or scored decisions with probabilities; the optimizer also works with your own backend.
From a source checkout with Python 3.10 or newer: python -m pip install ".[pandas,hub]". That installs the client; inference needs a running JEV-9B service or a GPU deployment documented in docs/deployment.md.
On one RTX 4090: up to 11.7x throughput over original order with long shared context, and 3.28x over NDV sorting with correlated fields. Both are reproducible synthetic workloads run by the project.
Inference through the bundled backend does; the benchmark ran JEV-9B on a single RTX 4090, and the digest does not list minimum hardware. The layout optimizer itself is independent of the backend.
Structured workloads with repeated policies, account context, catalog attributes, or correlated fields: per-row semantic judgments over large relational tables on a decision model.
0814wdwd's MIT-licensed Python repo: 701 stars, 9 forks, created 2026-09-29, last pushed 2026-10-08. Docs cover inputs, benchmarking, profiling, deployment, and a vLLM Open-Jev adapter.
More Jev resources like this
HTTP 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 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/deck31bPythonAn open-source, non-autoregressive System One decision engine inspired by TypeSafe AI: it evaluates state context against typed questions in a single forward pass, returning probability distributions and calibrated confidence without generating text.
Jev Repos & SDKs#community#github#python
Community
Atulmishra22/jev-litePython
Atulmishra22/jev-litePythonA Python library that turns Jev-style decision models into ordinary function calls: check returns a bool, classify picks one option, label picks every option that applies, and score rates on a scale, with probabilities behind every answer.
Jev Repos & SDKs#community#github#python
Community
jlowin/vibecheckPython
jlowin/vibecheckPythonA Bangla-focused adaptation of Laya, the non-autoregressive typed-decision model: targeted data preparation, fine-tuning and evaluation for classification, routing and scoring on Bangla, Banglish and code-switched text, with the checkpoint still to come.
Jev Repos & SDKs#community#github#python
Community
shamspias/laya-banglaPython
shamspias/laya-banglaPythonFrom 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.