Jev directory
CommunityJev Models & Reimplementations
Ru-Jev v1
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.
- Reading time
- 2 min
- Published by
- Community
- Added
- Nov 2026
Highlights
- Returns a calibrated probability over candidate options and picks one; it does not generate text.
- Covers three decision shapes: yes/no, choice from options, and ordinal scale.
- Compact mmBERT-base encoder with 307M parameters that runs on CPU.
- Held-out development split of 17,501 items: ECE 0.0054, temperature 1.82, micro accuracy 0.824.
- Trained from scratch on open, human-labelled datasets; not trained on Jev outputs.
Questions about this entry
Short answers drawn from the linked project or page.
Load the tokenizer and backbone with transformers AutoTokenizer and AutoModel, attach a linear head loaded from head.pt via hf_hub_download, and apply the temperature value from calibration.json. The model card includes the full usage snippet.
MIT, according to the repository metadata. The model is public and not gated, and it is listed as compatible with Inference Endpoints.
It is Russian-first, with English also listed among its language tags. It is trained on open, human-labelled datasets rather than on Jev outputs.
No. The model card states it is not affiliated with TypeSafe AI and not trained on Jev outputs; the name is a play on the category, not a claim of compatibility or endorsement.
On a held-out development split of 17,501 items it reports micro accuracy 0.824 and an expected calibration error of 0.0054. Per-task accuracy ranges from 0.964 on toxicity down to 0.490 on rwsd coreference.
Reasoning-heavy tasks are the weak spots for an encoder of this size: rwsd coreference 0.490, rcb 3-way NLI 0.518, and parus choice 0.600. The card says these are the focus of the next version.
No. It is a typed-decision model: given a context and a set of candidate options it returns a calibrated probability over the options and picks one, rather than producing free-form output.
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
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
A trained example bank for @ruvector/typesafe on Hugging Face, covering 150 CLINC150 intents across 10 domains plus 1,000 out-of-scope utterances. It stores labelled examples and frozen splits, not weights; heads and temperature calibration are refit on load.
Jev Models & Reimplementations#open-models#classification#local
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.