ClinicalJev
Open clinical text classifiers (0.8B-9B on Qwen3.5) with labels set at inference time, run locally and compared with Jev 1.13.
Overview
ClinicalJev is a family of open models for general-purpose clinical text classification. You define the task and candidate labels at inference time and use the same weights across tasks, getting probabilities over the predefined candidates without task-specific fine-tuning. It runs locally so sensitive healthcare data stays inside your facility.
Preview models come in 0.8B, 2B, 4B and 9B sizes built on Qwen3.5, with both LoRA and full weights on Hugging Face. Training covered English and Simplified Chinese only; other languages are not validated. The README includes benchmark tables comparing each size with Jev 1.13.0 on several clinical datasets.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which candidate label fits this clinical text? | labels defined at inference time |
How it works
- Define the clinical task and candidate labels
- Run ClinicalJev locally
- Get probabilities over candidates
- Apply thresholds in your system
Strengths
- Data stays on premises
- No per-task fine-tuning
- Several sizes for different hardware
Caveats
- v0.1 preview
- Validated only in English and Simplified Chinese
- Not a medical device
Getting started
Download a ClinicalJev model from Hugging Face (xinyuzhou/ClinicalJev-*).At a glance
- Type
- Project
- Status
- research
- Language
- Python
- Runtime
- local model
- Has code
- yes
Use cases and integrations
How curated lists describe it
An open-source general-purpose clinical Jev.
hellogumbo/awesome-jev
Frequently asked questions
Does data leave my facility?
No, it runs locally.
Which languages?
English and Simplified Chinese have been validated.
Is it compared with Jev?
Yes, benchmark tables include Jev 1.13.0.
Related projects
Mangaba AI's local, calibrated typed-decision model for Portuguese text, trained to optimize the Brier score; proprietary since v5.
ProjectproductionPython- blink★ 22
Open one-pass typed-decision model with a dependency-free C99 runtime that also builds to a 66 KB WebAssembly module.
ServicebetaC - Foq★ 11
Local open alternative to Jev using a 1.58-bit ternary 8B model to answer booleans, choices, scores and Pydantic extraction in ~25 ms.
ProjectbetaPython Runs the open Laya decision model entirely in a browser tab with WebGPU and ONNX Runtime Web: no server, no API key.
ServicebetaTypeScriptllama.cpp-based C++ server that turns any LLM into a System One decision model, reading answers from next-token logits.
ProjectbetaC++Three open 4B decision models for English, Turkish and German that serve the Jev /v1/systemone API on a CPU in about 3 GB of RAM.
ServicebetaPython