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.
User repository on GitHub
Open visual probability research: encode an image once, answer multiple structured questions. Public data, trained weights, reproducible baselines and failure cases.
Repo stats from the GitHub API, cached Oct 2026.
Encode an image once. Answer several questions from a shared probability distribution. OpenJev-Vision is an open research toolkit for visual probabilistic decisions. It includes original synthetic scenes, public-image experiments, downloadable trained readouts, and reproducible evaluations. The original OpenJev text-decision experiments remain available in this repository. Vision guide · Vision weights · Dataset · Vision results · 中文.
Quoted from the project's README. Read the full README.
Watch out
Experimental research release unaffiliated with TypeSafe; calibration is specific to the measured mixture and timings are single-machine M3 Pro observations. Copy reviewed against the repository README on 2026-10-04, not from running it.
TypeLLM/TypeLLMPython
hunkim/solar-mini4-jevPython
FeiLiuEM/open-medical-jevPython
mmastrac/djevPythonOriginal write-ups on what this entry is for.
Posts from builders shipping with Jev right now.
