Jev directory
CommunityJev Models & Reimplementations
Medical Jev 4B
LoRA adapter for Qwen3.5-4B that scores medical multiple-choice options in a single forward pass and returns a short JSON object with the chosen answer key and per-option probabilities, without explanations or chain-of-thought.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Built on Qwen3.5-4B as a LoRA adapter trained on medical multiple-choice items.
- Returns a short JSON object with the selected answer key and per-option probabilities.
- Accepts 2 to 52 candidate options and up to 8,192 tokens of input per item.
- Ships a ~130 MB adapter plus inference code; the ~9.3 GB base model downloads on first load.
- Adapter uses the custom jev-lora-v1 format and is not compatible with PEFT or chat generation.
Questions about this entry
Short answers drawn from the linked project or page.
Use Python 3.12 with PyTorch 2.8.0 built for CUDA 12.8. Download the bundle with hf download JuneYao/medical-jev-4b --revision v1.1.0, install its requirements.txt, then run predict_choices.py with --bundle pointing at the download and --input at a JSONL file. The setup was verified on Linux with an RTX 5090 32 GB.
Each call returns a short JSON object containing the item id, a status field, the answer key chosen from the supplied options, and a probabilities map giving a probability for every option. It does not generate explanations or chain-of-thought text.
Input is JSONL with one item per line. Each line has an id, a request.state holding the case material, and a request.questions.decision with instructions and a criteria map from option keys to option text. Each item has exactly one decision, and no ground-truth answer is required.
The released bundle supports 2 to 52 candidate options per item and up to 8,192 tokens of full input. That length covers the case material, the candidate options, and the chat template.
No. The adapter uses a custom jev-lora-v1 format and is not compatible with PEFT or chat generation interfaces. Load it with the bundled predict_choices.py script or the ChoicePredictor class instead.
The model card lists an MIT license and covers Chinese and English. The published examples and scenario descriptions are written in Chinese.
The card positions it for offline prototyping, teaching drills, and model evaluation, where results can be counted by a program or reviewed by a person. The published examples use constructed material and are described as illustrations of the calling convention, not as a medical capability evaluation.
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
Russian-first open typed-decision model on Hugging Face. Given a context and a set of candidate options it returns a calibrated probability and picks one, covering yes/no, choice, and ordinal scale decisions. Built on mmBERT-base with 307M parameters under an MIT license.
Jev Models & Reimplementations#open-models#calibration#classification
Community
Full-parameter fine-tune of Qwen3.5-4B for medical multiple-choice scoring. Given a case and candidate options, it returns the chosen option key with per-option probabilities as short JSON, without explanations or chain-of-thought.
Jev Models & Reimplementations#open-models#classification#python
Community
Satirical Python script that answers a query with DO or DO_NOT plus a random confidence percentage, presented as a natural alternative to Jev. MIT-licensed, Python 3.8+, one dependency, no API keys or model weights.
Jev Models & Reimplementations#github#python#open-models
CommunityAhmad170412/Koin
Hugging Face model repo hosting llm-as-jev, a fine-tune of Qwen/Qwen3-0.6B tagged for decision-model and multiple-choice work. Ships safetensors weights, a transformers configuration, and an Apache-2.0 license.
Jev Models & Reimplementations#open-models#local#classification
Community
A JEV decision model built on inclusionAI/Ling-3.0-tiny with AReno, shipping a decision head and rank-128 backbone LoRA adapters. Serves choice, noul, and score questions over an areno /v1/systemone endpoint on Apple Silicon MLX and CUDA.
Jev Models & Reimplementations#open-models#local#python
Community
A 561M-parameter XLM-RoBERTa-large decision model given one extra training epoch on Jev's answers, with about 23% of examples black-box distilled from Jev's public API. Supports noul, choice, and score questions in 18 languages and loads through the bundled naluri.py.
Jev Models & Reimplementations#open-models#calibration#benchmark
Community
From 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.