User repository on GitHub
A family of System One-style models fine-tuned from Qwen3.5, designed for one-pass typed decisions with calibrated probabilities.
- Forks
- 51
- Last pushed
- Updated Oct 2026
Repo stats from the GitHub API, cached Oct 2026.
Jev directory
CommunityJev Practices & Patterns
Decider
An independent open reproduction of the System One model class: Qwen3.5-based 2B and 35B mixture-of-experts models that return typed Choice, Score, and Noul probabilities in one forward pass, with nothing distilled from Jev.
- Stars
- 1.2k
- Language
- Python
- License
- APACHE-2.0
- Reading time
- 3 min
- Published by
- Community
- Author
- Mapika
- Last verified
- Sep 2026
About this project
A language model that does not generate text. It reads a state and a set of typed questions and returns, from one forward pass, a probability distribution for every question. A typed decision is a question with a fixed answer set: Choice over 2 to 255 options, Score over 2 to 10 described levels, or Noul, the probability of yes. There is no decoding, no parsing, and no output outside the options you defined.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
Highlights
- decider-2b is built on Qwen3.5-2B-Base and decider-35b-a3b on Qwen3.5-35B-A3B-Base, trained on public data plus labels from a local Qwen3.5-27B teacher.
- The project says nothing was distilled from Jev and it is not affiliated with or endorsed by TypeSafe AI.
- On JevBench it ranked #10 of 36 (decider-35b-a3b, 68.9) and #21 (decider-2b, 64.6); on the Decision Index the NVFP4 build sits 4th of 32.
- decider-2b v10 adds 384 steps of calibration-aware RL, lifting sampled browser play from 83% to 93% and belief from 0.47 to 0.22 nats above the exact laws.
- Published on PyPI as decider-ai with MPS acceleration for Apple Silicon; Pong decisions run in 43 ms median on one B300 in bf16.
Quickstart
bash
pip install decider-ai
git clone https://github.com/Mapika/decider && pip install -e ".[serve]"Watch out
Apache-2.0. Unaffiliated with TypeSafe; serving needs Python plus a GPU or Apple Silicon MPS, the 2B trails on calibration and the 35B on cost, and a Blackwell cached shared-state bug was fixed in 1.0.2.
Questions about this entry
Short answers drawn from the linked project or page.
Independent, and the project is explicit: nothing was distilled from Jev and it is not affiliated with or endorsed by TypeSafe AI. The models are built on Qwen3.5-2B-Base, Qwen3.5-4B-Base and Qwen3.5-35B-A3B-Base, trained on public datasets plus labels from a local Qwen3.5-27B teacher.
The README tabulates third-party boards it did not run itself. On JevBench v1.5.2, Jev 1.13.0 sits at 72.1 in third place, decider-4b v2 at 71.1 in seventh, decider-2b at 45.1 in twenty-eighth and decider-35b-a3b at 27.5 in forty-first. On the Decision Index, Jev is listed at 57.91 against 57.33 for decider-chat-gemma4-31b in second. Reruns move these ranks.
GGUF Q4_K_M builds are 1.3 GB for the 2B and 2.7 GB for the 4B, served through llama.cpp on CPU, CUDA or Metal. The README reports 0.12 to 0.31 s for a 40 to 120 token request with the 2B on eight server CPU threads. The PyTorch path falls back CUDA, then Apple MPS, then CPU.
The README says decider-4b v2 is held back on Intelligence by the sealed yes/no items (43 open against 19 sealed, chance-corrected) and that the 35B is gated by cost. Recent entries are fixes rather than wins: a Blackwell cached shared-state bug in 1.0.2, and a shared-prefix memory bound in 1.8.1 that took a 32-document retrieval request from out-of-memory to a 68.5 GB peak.
decider-4b, at 8.4 GB in bf16, with held-out accuracy 0.784 on the regression set and a JevBench public hard tier of 0.649.
Install decider-ai[gguf] and load a GGUF checkpoint through llama.cpp; on eight server CPU threads a 40 to 120 token request takes 0.12 to 0.31 s with the 2B and 0.3 to 0.7 s with the 4B.
The PyPI name is decider-ai but the import name is decider, so you pip install decider-ai and then from decider.infer import Decider.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Standard One SH (StandardThinking)
LinkedIn: What multimodal decision models actually change, and what they do not.
LinkedInJevDirectory.org
Hacker News: Decision Index 0.2.1 scores open reproductions of Jev on a frozen suite.
jackbravo4 pointsDecision Index 0.2.1 – 70 open reproductions of TypeSafe Jev's Decision ModelRead the thread on Hacker NewsHacker News: Pointer to a DiffusionGemma plus vLLM tutorial for a Jev-shaped classifier.
1 pointHow to build a Jev-style classifier with DiffusionGemma and vLLMRead the thread on Hacker NewsMore Jev resources like this
Apache-licensed, locally runnable Jev-style decision models (0.8B, 4B, 9B on Qwen3.5) with released weights, training code, a System One-compatible server, frozen eval suites, and a playground.
Jev Practices & Patterns#community#python#open-models
Community
jaredpalmer/kevPython
jaredpalmer/kevPythonA from-scratch one-pass option scorer that maps a context and a changing list of text options to one probability per option, adding a hashed n-gram encoder, rival-aware attention, a gated head, and temperature scaling; JSONL data is drop-in compatible with jevlike.
Jev Practices & Patterns#community#python#calibration
Community
olanotolu/jevbetterPython
olanotolu/jevbetterPythonAn independent baseline that reads typed option probabilities straight from a frozen Qwen3.5-4B's logits in one forward pass, reproducing Jev's interface pattern with open models rather than Jev's undisclosed model or training.
Jev Practices & Patterns#community#python#open-models
Community
TheoLeeCJ/SemIfPython
TheoLeeCJ/SemIfPythonA ModernBERT-based open decision model that evaluates Choice, Score, and Noul schemas in one non-autoregressive pass; Verdict 2.0 reports 77.10% accuracy on 2,000 held-out enterprise decisions.
Jev Practices & Patterns#community#python#open-models
A self-hosted implementation of TypeSafe's Jev System One API powered by the 400M-parameter GLiFormer encoder: it serves choice, score, and noul and drops into the official typesafe-sdk via TYPESAFE_BASE_URL, but trails Jev on reasoning-heavy tasks.
Jev Practices & Patterns#community#python#open-models
Community
logan-markewich/jeffPython
logan-markewich/jeffPythonAn open re-creation of Jev's decision model on an open-weights Qwen3.5-4B: it answers a list of typed questions in one forward pass over POST /v1/systemone in about 200 ms, never writing free text.
Jev Practices & Patterns#community#python#open-models
Community
kshetrajna12/reflexPython
kshetrajna12/reflexPythonFrom 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.