Jev directory
CommunityJev Models & Reimplementations
Kieve-JeV Fractal 300M
Encoder-based typed decision engine from ScaiResearch Labs: a frozen ModernBERT-base backbone with rank-16 LoRA adapters that returns one answer per typed question as a calibrated distribution over the supplied options. Ships a training checkpoint and Kieve inference config.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Takes a state of any text plus typed questions and returns a calibrated distribution over your supplied options.
- Never invents options: the distribution is defined only over the options passed in.
- Backbone is answerdotai/ModernBERT-base, frozen, with LoRA adapters of rank 16.
- Total parameters are 308.9M, with a jev variant using state and question encoders plus a paired cross-attention option scorer.
- Reported held-out JvD metrics: aegis2 0.449 and atbench500 0.525.
Questions about this entry
Short answers drawn from the linked project or page.
It is an encoder-based typed decision engine. It takes a state (any text) and a set of typed questions and returns, for every question, one answer with a calibrated probability distribution over the supplied options.
The backbone is answerdotai/ModernBERT-base, kept frozen, with LoRA adapters of rank 16. The jev variant adds state and question encoders, a paired cross-attention option scorer and a separate noul head.
Total parameters are 308.9M. The release includes checkpoint_latest.pt, whose model_state holds the weights and a calibrated softmax temperature, plus config.json, the inference config consumed by Kieve.
No. The model card states the model never invents options and that the distribution is defined over the options you pass in. A separate noul head returns a probability for the yes/no noul primitive instead.
On held-out JvD the card reports aegis2 0.449 and atbench500 0.525. Decision Index 0.2.1 records decision_index 1.38 and raw 21.40 over 150,759 requests, marked complete.
The model is licensed Apache-2.0. The ModernBERT backbone is subject to its own license, which applies separately from the model's Apache-2.0 terms.
No. The Hugging Face page shows no Inference Provider deployment and invites users to ask for provider support, so the checkpoint and config are intended to be run through your own Kieve runtime.
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
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
A 46M-parameter Apache-2.0 decision model on Hugging Face, fine-tuned from q-project/N-Q-26M-JEV for typed classification. It exposes the custom NQ4ForDecision class to Transformers and lists Russian and English, plus NLI, sentiment, news, and toxicity datasets.
Jev Models & Reimplementations#open-models#classification#calibration
Community
MLX builds of Jev-Style-2B-Decision-v4 for Apple silicon: bf16 (3.76 GB) and 8-bit (2.00 GB) folders for text and JSON states, run through one shared jev_style_decision_mlx.py runtime under Apache-2.0.
Jev Models & Reimplementations#open-models#calibration#local
Community
A 26M-class decision model on Hugging Face, fine-tuned from q-project/N-Q-26M-Chat. It ships a custom NQ4ForDecision class for text classification and feature extraction, is tagged calibrated and decision-model, and publishes Apache-2.0 weights with Russian and English support.
Jev Models & Reimplementations#open-models#calibration#classification
Community
An EmbeddingGemma 2 fine-tune that answers choice, score, and noul questions from a state and a plain-data schema, with a 0.5M-parameter schema-conditioned scorer over 768-d embeddings.
Jev Models & Reimplementations#community#open-models#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.