from preceding post in gpt-4 thread<br><div><br></div><div><a href="https://github.com/karpathy/llama2.c" target="_blank">https://github.com/karpathy/<wbr>llama2.c</a> was purportedly a 700-line pure C llama 2 simplified llama.cpp example implementation, demo of it ported to an esp32 at <a href="https://github.com/DaveBben/esp32-llm" target="_blank">https://github.com/<wbr>DaveBben/esp32-llm</a> pic attached repo has an animated gif</div><div><br></div><div>the project may have been unmaintained for 2 years and many PRs and issues have accumulated. calling out among contributors for a maintainer might change that.</div><div><br></div><div>gemini said riscv is where it&#39;s at now and people are making distributed AI networks with small chips on reddit. i didn&#39;t synthesize that information i&#39;m afraid.</div><div><br></div><div>the post also mentioned a palm-sized programmable robot dog at <a href="https://www.petoi.com/products/petoi-bittle-robot-dog">https://www.petoi.com/products/petoi-bittle-robot-dog</a> -- it looks more like twice as big as the palm standing up. there are hopefully 3d print projects for things like this around nowadays.</div><div><br></div><div>and a raspberry pi edge AI assistant at <a href="https://github.com/Chappie02/Multi-Modal-AI-Assistant-on-Raspberry-Pi-5">https://github.com/Chappie02/Multi-Modal-AI-Assistant-on-Raspberry-Pi-5</a> pic also attached!</div>