User repository on GitHub
Qwev: A Training-Free, Qwen-Based Jev
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Models & Reimplementations
Qwev
A training-free runtime that turns dense Qwen3 or Qwen3.5 checkpoints into a Jev-style Noul, Choice and Score service, sharing one state prefill across isolated questions and reading each answer from candidate-token logits.
- Stars
- 3
- Language
- Python
- License
- MIT
- Published by
- Community
- Last verified
- Oct 2026
About this project
TypeSafe Jev presents a useful interface for decision workloads: provide a state once, attach many typed questions, and return all decisions without making an autoregressive model write a JSON response token by token. Each question declares its answer space as a binary probability, a set of named choices, or a numeric scale.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
- Languages
- Python
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- Included 27-question Qwen3.5-9B fixture on an A100: 0.500 s direct scoring versus 13.554 s generating the same JSON (27/27 labels).
- On 337 public TypeSafe showcase questions, direct scoring matched 279 labels; JSON generation matched 292 in 237.58 s.
- Execution modes: batch and shared for Qwen3.5; packed and cascade additionally for Qwen3, reaching 0.212 s on Qwen3-8B.
- POST /v1/systemone takes one state with up to 32 questions; the matrix endpoint covers up to 64 states by 32 questions.
Watch out
It adopts Jev's public problem framing but not its architecture or weights, and softmax values are not calibrated correctness probabilities. MIT licensed; rejects MoE and quantized checkpoints.
Questions about this entry
Short answers drawn from the linked project or page.
No. It is training-free: the runtime keeps the model weights unchanged and reads allowed answers directly from candidate-token logits of dense Qwen3 and Qwen3.5 checkpoints.
The shared state is prefilled once and copied into independent question rows; each question reads the state and its own prompt but never another question's prompt or answer.
POST /v1/systemone takes one state with up to 32 questions; the matrix endpoint handles multiple states by multiple questions, up to 64 states, 32 questions and 256 state/question pairs in one request.
They are softmax values over the allowed answer tokens, not calibrated probabilities that the answer is correct, so treat them as relative confidence rather than a guarantee.
It rejects MoE, quantized, sliding-window and custom-RoPE checkpoints, and Qwen3.5 cannot use the packed or cascade modes because its recurrent Gated DeltaNet layers do not obey a block attention mask.
Install the package with pip install -e ., then start the factory app with uvicorn qwev.api:create_app --factory; it exposes /health, /v1/systemone and /v1/systemone/matrix on the chosen host and port.
It reads the architecture from config.json rather than a fixed model list, but only Qwen3-8B and Qwen3.5-9B ship checked-in real-weight reports, so validate isolation, memory, accuracy and latency before relying on another checkpoint.
More Jev resources like this
10GitHub stars
A local Jev-compatible server that reads Choice, Score and Noul answers from the logits of any MLX checkpoint or OpenAI-compatible endpoint with no training, serving POST /v1/systemone as a drop-in for the official TypeSafe SDK.
Jev Models & Reimplementations#community#mlx#calibration
Community
bradAGI/rulingPython
bradAGI/rulingPython877GitHub stars
A local server that reads next-token probabilities from a fine-tuned Spark model and returns Choice, Score, and Noul answers with zero generated tokens.
Jev Models & Reimplementations#community#python#open-models
Community
Rizzo-AI-Academy/rizzo-flowPython
Rizzo-AI-Academy/rizzo-flowPython130GitHub stars
An open recipe for a 27B multimodal decision model trained with full-weight SFT on 73,000 examples over one H200, serving choice, noul and score on /v1/systemone with per-checkpoint provenance and released calibration plots.
Jev Models & Reimplementations#community#calibration#qwen
Community
denis-pplx/autojevPython
denis-pplx/autojevPython17GitHub stars
Jev-compatible decision models with vision input: LoRA fine-tunes of Qwen3.5 at 4B and 9B serving Choice, Score and Noul on the /v1/systemone wire format, with screenshots and photos placed beside text in the state. Runs on your own NVIDIA GPU; weights are non-commercial.
Jev Models & Reimplementations#community#vision#open-weights
Community
Xiaooolong/vevPython
Xiaooolong/vevPython77GitHub stars
An Apache-2.0 0.8B decision model that returns calibrated probabilities for yes/no, choice and score questions on CPU. Distributed as GGUF for llama.cpp and served behind a Jev-style /v1/systemone API, so existing Jev clients mostly need only a new base URL.
Jev Models & Reimplementations#community#gguf#local-inference
Community
kouhxp/gutsyHTML
kouhxp/gutsyHTML58GitHub stars
An Apache-2.0 decision engine with a TypeSafe-compatible endpoint, document answers with source passages, and a ViZDoom demo that picks actions from structured game state.
Jev Models & Reimplementations#community#python#local
Community
deepanwadhwa/OpenDecisionPython
deepanwadhwa/OpenDecisionPythonFrom the Jev guides
Original write-ups on what this entry is for.
Jev takes text only; Clef-omni adds image, audio, and video input and Liquid d1 adds images, each keeping the same typed-answer shape. Where that matters, where it does not, and the costs the vendors underplay.
Jev Models & Reimplementations#multimodal#open-models#decision-model
Read article
One model in September, a dozen by October. Clef, Liquid d1, Perplexity, and OpenAI all ship the same typed-answer shape now. How to read the rival benchmarks and pick on your own task.
Jev Models & Reimplementations#decision-model#benchmarks#pricing
Read article
Hundreds of repositories carry the system-one tag, but the ecosystem is a dozen ideas in different sizes. A tour of the families, the licenses, and what a download does not include.
Jev Models & Reimplementations#open-models#hugging-face#local
Read article
From the Jev community
Posts from builders shipping with Jev right now.