⬢github C++ · 80 ★ +23 since we first saw it · pushed 2 d ago · MIT
Low-Zi-Hong/ESP32s3-LLM-Cluster
7-node ESP32-S3 cluster running a 0.4B LLM via 1.58-bit (BitNet) ternary quantization over SPI daisy-chain
A hobbyist hardware project that runs a ~0.5B-parameter LLM split across a cluster of seven ESP32-S3 microcontroller boards. One master board handles the tokenizer and embeddings; six compute nodes each execute several transformer layers using 1.58-bit BitNet ternary quantization, passing hidden states along a high-speed SPI daisy-chain to generate text.
Why now: It was featured on Hacker News, likely because running a transformer LLM on cheap microcontrollers via BitNet quantization is a surprising and impressive demo.
Who it is for: Embedded and hardware hobbyists curious about extreme model quantization and running neural networks on microcontrollers.
Stars over our 14 snapshots: 57 to 80, since 3 h ago.
Where people talked about it
API: https://socialmediatrends-api.osmike.com/v1/repos/Low-Zi-Hong/ESP32s3-LLM-Cluster