User repository on GitHub
SDK-first, independently maintained SemIf fork for fast, self-hosted semantic decisions.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Repos & SDKs
FastJev
Self-hosted Python SDK and System One-compatible API for runtime-defined Choice, Boolean and Score decisions on pinned open models, across Torch, vLLM, MLX, llama.cpp and WebGPU. Each result records probabilities, token usage, timing, model revision and prompt version, with committed row-level benchmarks.
- Stars
- 21
- Language
- Python
- License
- MIT
- Published by
- Community
- Last verified
- Oct 2026
About this project
Deploy an open source Jev implementation on your own infrastructure. FastJev is an open source implementation of Jev for deployment on infrastructure you control. It continues SemIf as an independently maintained fork. Pinned open models run the Choice, Boolean, and Score interface through Torch, vLLM, MLX, llama.cpp, and optional EXL3 quantization. The WebGPU demo provides browser local inference.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
- Languages
- Python
- Capabilities
- Open sourceSelf-hostedRuns locally
Highlights
- Validates 2 to 16 options and returns typed results with probabilities, usage, timing and provenance.
- RTX 5090 comparison: vLLM 82.73ms median for three decisions vs Torch 161.97ms and llama.cpp 2.344s.
- Qwen3.5-4B scored 0.813 on 144 authored rows and 0.845 agreement on a TypeSafe public subset.
- fastjev-score runs JSONL jobs and fastjev-serve exposes POST /v1/systemone and GET /v1/models.
Watch out
MIT licensed. It is an independent SemIf fork, not an official TypeSafe SDK, and results stay calibrated=False unless an identity-bound TemperatureCalibration is attached; model weights stay on upstream hosts under their own licenses.
Questions about this entry
Short answers drawn from the linked project or page.
No, it runs pinned open checkpoints locally through Torch, vLLM, MLX, llama.cpp or EXL3 and exposes a System One-compatible HTTP API. A published subset comparison reached 0.845 modal agreement against Jev's 0.883.
The README recommends Qwen3.5-4B at a pinned revision for highest measured quality, with 0.813 authored balanced accuracy. MiniCPM5-2B is described as the size/quality balance at 0.686, and Qwen3-0.6B is the smallest starting point at 0.440.
Typed results with probabilities, token usage, timing, the model revision and the prompt version. Torch, MLX, llama.cpp and EXL3 return results in one scoring pass with zero output tokens, and Choice results validate that the options number between 2 and 16.
On an RTX 5090 comparison, vLLM posted a median of 82.73 ms for three decisions against Torch at 161.97 ms and llama.cpp at 2.344 s. Qwen3.5-4B needs about 9 GB of disk for the source checkpoint and measured 7.891 GiB of peak allocated GPU memory in the historical SDK smoke run.
pip install 'fastjev[torch]' for the resident Torch backend, or the llama-cpp extra to run GGUF files from disk or Hugging Face. `fastjev-serve` exposes POST /v1/systemone and GET /v1/models, and `fastjev-score` runs JSONL jobs. The default backend wants Python 3.10+ and either exactly one visible CUDA GPU or Apple Silicon MPS.
Results stay calibrated=False unless an identity-bound TemperatureCalibration is attached, so the raw numbers are not calibrated confidence. The code is MIT and it is an independent SemIf fork rather than an official TypeSafe SDK, and the model weights stay on upstream hosts under their own licenses.
Yes, through decide_batch, which applies the same questions to multiple inputs and retains input order; batching is opt-in since the default batch_size is 1. The size counts decision prompts, not states, and batch shape can change reduced-precision probabilities and close decisions. It is a Python SDK feature, so the CLI and HTTP service do not batch automatically.
More Jev resources like this
130GitHub 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/autojevPython109GitHub stars
Open decision models from a self-improvement loop that publishes every experiment, failures included. They answer Noul, Choice and Score questions on the same POST /v1/systemone wire format as Jev, returning a calibrated probability per option in one forward pass; the 27B scores 65.64 on the Decision Index 0.3 set.
Jev Models & Reimplementations#community#decision-models#system-one
Community
Shanghua-Gao/RSI-JevHTML
Shanghua-Gao/RSI-JevHTML10GitHub 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/rulingPython4GitHub stars
A pre-release Python framework for predictions on mixed text and tabular data: Jev Noul and Score questions turn text columns into numeric features, then a tabular foundation model such as TabPFN learns from labeled rows in context. Offers hosted and fully local backends.
Jev Repos & SDKs#community#tabular#predictions
Community
edamame-labs/tab-jevPython
edamame-labs/tab-jevPython2GitHub stars
Ruby gem that makes System One measurement a language primitive: psi prepares a state, judge/choose/score return calibrated distributions, and ? collapses them for control flow. Ships TypeSafe, protocol (/v1/systemone) and CUA providers behind one interface with its own spec suite.
Jev Repos & SDKs#community#ruby-gem#sdk
Community
innocentdiaz/s1_rubyRuby
innocentdiaz/s1_rubyRuby46.7kGitHub stars
Privacy-first, self-hosted personal knowledge management system with block-level references and a local-first Markdown editor; the project integrates Jev's decision model for high-speed document classification, flashcard intent categorization and automated tag routing.
Jev Integrations & Adapters#community#note-taking#knowledge-base
Community
siyuan-note/siyuanTypeScript
siyuan-note/siyuanTypeScriptFrom the Jev guides
Original write-ups on what this entry is for.
Closed weights have not stopped the ecosystem from rebuilding the shape in the open. API-compatible servers, retrained heads on small models, and inference-time tricks that skip generation entirely.
Jev Repos & SDKs#open-models#research#local
Read article
From the Jev community
Posts from builders shipping with Jev right now.