Home /Open models and replicas /jev-visual

jev-visual

Educational visual-inference variant on Apple Silicon: shared context, direct candidate scoring.

From the readme

Jev Visual English · 简体中文 A small, runnable project for learning vision-language model inference on Apple Silicon. Use Qwen3.5-0.8B with MLX to answer multiple questions about one image: choose an option, judge yes/no, or score ordered levels. Includes a local browser UI, CLI and HTTP API. This project explores a Jev-like inference pattern for open …

Details

Section
Open models and replicas
owner
hr98w
stars
162
forks
20
pushed
2026-09-18
Language
Python
License
MIT

What it is

Form
CLI
Host agent
Standalone
Audience
Researchers
Maturity
Docs
mocked or unclear Jev call

Useful for

Best intent matches