Uses an iPhone connected over USB-C to speed up prefill and extend context for a 27B Qwen model running locally on a Mac.
hr98w/jev-visual
A small educational project for running vision-language model inference on Apple Silicon with MLX, scoring candidate answers about one image.
- Звёзды
- 319
- Форки
- 0
- Открытые issues
- 0
- Последний коммит
- 28 дней назад
Ключевые возможности
- 01Asks several questions about one image in a single shared-context pass.
- 02Scores candidate answers directly from model logits instead of generating JSON.
- 03Provides a local browser UI, CLI and HTTP API.
- 04Includes visual demos such as a sorting game, Breakout variant and camera gesture recognition.
- 05Ships benchmarks comparing generation-based and shared-prefix scoring.
Сценарии использования
- Learning how vision-language models can be used for classification without structured generation.
- Experimenting with multi-question image analysis on a local Apple Silicon Mac.
- Studying KV cache reuse and candidate scoring in MLX.
- Prototyping camera-based interactions with a small local model.
Темы
Частые вопросы
- Что такое jev-visual?
- A small educational project for running vision-language model inference on Apple Silicon with MLX, scoring candidate answers about one image.
- jev-visual можно использовать бесплатно? Какая у него лицензия?
- jev-visual распространяется с открытым исходным кодом под лицензией MIT. Перед использованием в коммерческих продуктах проверьте условия лицензии.
- jev-visual всё ещё поддерживается?
- Активно поддерживается. Последний коммит в jev-visual — 21 сент. 2026 г..
- Для чего можно использовать jev-visual?
- Learning how vision-language models can be used for classification without structured generation. Experimenting with multi-question image analysis on a local Apple Silicon Mac. Studying KV cache reuse and candidate scoring in MLX. Prototyping camera-based interactions with a small local model.
- Где скачать jev-visual?
- Исходный код и инструкции по установке — на GitHub: https://github.com/hr98w/jev-visual
Описание создано ИИ на основе README проекта. Статистика GitHub обновляется ежедневно.
Похожие проекты
Swift and Metal runtime that runs the 26B Gemma 4 26B-A4B model on Apple Silicon Macs with about 2 GB of RAM by streaming experts from SSD.
Open-source native unified multimodal models from SenseNova that handle image understanding, text-to-image generation, and image editing in one architecture.
Runs Claude Code against open AI models on your Apple Silicon Mac via an MLX-based Anthropic API server, with no cloud model or API key.
A 9B-parameter open-weight multimodal model for visual understanding, spatial reasoning, tool use, and embodied-AI research, built on Qwen3.5 and C-RADIOv4-H.
Open-source native vision-language models (NEO series) that process pixels and words in one encoder-free dense architecture, with 2B and 9B releases.