This guide explains how to install StackFlow on a Raspberry Pi connected to an LLM-8850 Card. Before proceeding, complete the LLM-8850 Card environment and driver setup, and confirm that axcl-smi can detect the accelerator card.
Run the following command to check the Raspberry Pi OS codename:
lsb_release -cs The current repository supports Debian 12 (bookworm) and Debian 13 (trixie).
Install the GPG key:
sudo wget -qO /etc/apt/keyrings/StackFlow.gpg https://repo.llm.m5stack.com/m5stack-apt-repo/key/StackFlow.gpg For Debian 12 (bookworm):
echo 'deb [arch=arm64 signed-by=/etc/apt/keyrings/StackFlow.gpg] https://repo.llm.m5stack.com/m5stack-apt-repo bookworm llm8850' \
| sudo tee /etc/apt/sources.list.d/StackFlow.list > /dev/null For Debian 13 (trixie):
echo 'deb [arch=arm64 signed-by=/etc/apt/keyrings/StackFlow.gpg] https://repo.llm.m5stack.com/m5stack-apt-repo trixie llm8850' \
| sudo tee /etc/apt/sources.list.d/StackFlow.list > /dev/null Update the package index:
sudo apt update Install the base runtime first:
sudo apt install lib-llm llm-sys Then install the services required by your application. For example, install the large language model service and OpenAI-compatible API with:
sudo apt install llm-llm llm-openai-api Services also require corresponding model packages. Run the following command to list the packages available from the configured repository:
apt list 2>/dev/null | grep '^llm-' Model packages can occupy substantial storage. Refer to the model library and install only packages compatible with the LLM-8850 platform that your application requires.
After installation, continue with the LLM-8850 Card OpenAI API or browse other StackFlow development examples.