User repository on GitHub
- Forks
- 1
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Demos & Experiments
Nemotron_Jev
Self-hosted Decision Lab that asks typed choice, yes-no, and score questions through a browser explorer and a TypeSafe-shaped local API backed by a diffusion checkpoint.
- Stars
- 14
- Language
- Python
- License
- MIT
- Published by
- Community
- Author
- pst2154
- Last verified
- Oct 2026
About this project
Ask typed questions about text or JSON and inspect model-derived probability distributions in a browser. One container runs the model, the original Decision Lab explorer, and a TypeSafe-shaped API. This branch defaults to dense Nemotron-Labs-Diffusion-8B. It is not a mixture-of-experts model, Jev, an official TypeSafe service, or a calibrated replacement for Jev. The repository name describes its interface, not its model identity.
Quoted from the project's README. Read the full README.
Works forJev for Classification
Used inJev for Software & DevTools
- Languages
- Python
Highlights
- Choice, yes-no, and score questions with per-option probabilities plus a browser explorer and raw JSON.
- Single-token diffusion-head scoring over allowed candidates with no invented confidence values.
- Pinned checkpoint revisions with container images that fetch weights into a persistent volume.
- Classification-only API surface with no chat completions or multi-token generation.
Watch out
This branch serves a dense diffusion checkpoint, not Jev or an official TypeSafe service, and its probabilities are explicitly not empirically calibrated; the server has no auth, TLS, or rate limiting, so do not expose it directly. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
No. The repository states the name describes the interface, not the model identity: it is a dense diffusion checkpoint behind a compatible API, not Jev or a calibrated replacement.
They are measured in-repo procedures with pinned revisions, build flags, and logs, but they describe this container's own accuracy and latency on JevBench-style tasks, not parity with the hosted Jev service.
Do not. The server has no built-in authentication, TLS, or rate limiting, so keep it on loopback or behind an SSH tunnel or authenticated reverse proxy.
It is accepted only as a compatibility identifier: the returned model is the configured checkpoint name and the request never routes to TypeSafe, with no TypeSafe key needed.
Dense Nemotron-Labs-Diffusion-8B, pinned to revision 16c67f0560b912e93e0cabb6e0c4f5c3086d95fc. API responses and health report the configured model, while jev-latest remains an input compatibility alias and never routes to TypeSafe.
Docker with the NVIDIA Container Toolkit, a CUDA-13-compatible driver, and an H100/H200-class Hopper GPU; the prebuilt image targets compute capability 9.0, not B200/B300 or L40S, and H100 80 GB was tested while H200 was not.
For each question, options get tokenizer-verified single-token codes; the model causally encodes the prompt, then scores one masked answer position with its diffusion head, and a softmax over only the allowed candidate logits gives the distribution. Choice takes the highest probability, Noul the yes probability, and Score computes sum(index * probability).
More Jev resources like this
46GitHub stars
Open 150M encoder that reproduces the typed noul, choice, and score interface in one forward pass, trainable on a free GPU and runnable locally for nothing.
Jev Demos & Experiments#community#jevbest#catalog-import
941GitHub stars
Library that adds schema-guaranteed typed outputs to existing autoregressive LLMs without changing weights, served through SGLang or a hosted API.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
TypeLLM/TypeLLMPython
TypeLLM/TypeLLMPython45GitHub stars
Drop-in service that exposes Upstage Solar models through the System One shape, accepting state plus noul, choice, or score questions.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
hunkim/solar-mini4-jevPython
hunkim/solar-mini4-jevPython38GitHub stars
Research recipe that fuses two frozen open readers into Jev-style yes/no judgments with a fit-free router, confidence gate, and candidate sets.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
FeiLiuEM/open-medical-jevPython
FeiLiuEM/open-medical-jevPython62GitHub stars
Simulation study that drives one xArm7 apple-to-plate task with Jev, a large reasoning model, and a small model through identical incremental controls.
Jev Demos & Experiments#community#jevbest#catalog-import
40GitHub stars
Open research toolkit that encodes an image once and answers multiple typed choice, noul, and score questions from a shared distribution, plus a local 0.6B text-decision model.
Jev Demos & Experiments#community#jevbest#catalog-import
Community
IamBusy/OpenJev-VisionPython
IamBusy/OpenJev-VisionPythonFrom the Jev guides
Original write-ups on what this entry is for.
A game tick is a state plus a legal action list. How Clash Royale, Minecraft, and the earlier Mario, Pokémon, StarCraft, and drone loops ask Jev one step at a time.
Jev Demos & Experiments#gaming#agents#control
Read article
From the Jev community
Posts from builders shipping with Jev right now.