Jev directory
CommunityJev Models & Reimplementations
JEV Train Dataset (SII-liyang2024)
Hugging Face dataset repository holding four supervised fine-tuning configs for JEV decision models: AHAT and PARTNR next-action choices, plus AHAT and PARTNR progress choices over a fixed 21-option grid. Gated access; trainer records and audits only, no checkpoints.
- Reading time
- 2 min
- Published by
- Community
- Added
- Oct 2026
Highlights
- Four configs: ahat and partnr next-action decisions, plus ahat_progress and partnr_progress progress choices.
- Progress configs classify onto a fixed 21-point grid with option codes A through U covering 0.00 to 1.00.
- ahat_progress carries 513,140 train rows; partnr_progress carries 23,121 train rows.
- All four configs keep a custom calibration split alongside train, validation and test.
- The repository is gated: you must accept the conditions while logged in to access files and content.
Questions about this entry
Short answers drawn from the linked project or page.
Use the Hugging Face datasets library and name the config, for example load_dataset("SII-liyang2024/jev-train", "ahat_progress") or the same call with "partnr_progress". Each progress record contains messages, candidate_count (always 21) and target_index.
The ahat and partnr configs are next-action decision-choice data. The ahat_progress and partnr_progress configs classify progress on a fixed 21-point grid instead of picking a next action.
No. The repository is publicly visible, but you must log in or sign up and agree to share your Hugging Face information to access the files and content.
No. The progress targets are supervised categorical labels for candidate cross-entropy over 21 options, not scalar-head regression targets. Progress is quantized to the nearest 0.05 with midpoint-up rounding before it becomes a label.
No model checkpoint and no tokenizer are included. The uploaded trainer files contain the model input and categorical target only, plus metadata manifests and audits retained for provenance.
AHAT progress uses a scene-prefix input built from the scene graph plus raw plan actions, with the reference-plan position t/T quantized. It is a weak proxy for reference-plan progress and does not certify natural-language task completion.
ahat and ahat_progress each hold 513,140 train, 29,945 validation, 44,251 calibration and 28,397 test rows. partnr holds 5,585 / 583 / 279 / 193, and partnr_progress holds 23,121 / 2,842 / 1,614 / 1,318. Total repository file size is 28.8 GB.
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.