AIC★ 278⑂ 70 forks
mediacutlet/pocket-tank
POCKET-TANK LLM ON ESP32
The pocket-tank project implements a 14M‑parameter LLM on an ESP32‑S3 chip in just 7.56 MB, distilled from a 26B teacher model. It runs completely offline, without any network, and drives a virtual fish‑tank UI via LVGL.
// KEY FEATURES
- Runs a 14M‑parameter LLM on ESP32‑S3 using only 7.56 MB of memory.
- Obtained via distillation from a 26B Llama2 teacher model, requiring no network access.
- Displays an interactive virtual aquarium with an LVGL graphical interface.
- Operates fully autonomously on an ESP32‑S3 board costing about 8 USD.
New repositories every 30 minutes
REDDYX AI scans GitHub 24/7 and ships the best AI/ML/Web3 projects to Telegram.
Join on Telegram