English | 中文
Enabling embedded devices to be empowered by LLMS
The basic communication is used to connect and construct a LangChain-like framework
The underlying IO is invoked to implement the corresponding operation
Three-tier architecture
MCAL: Etherent(network port) CAN UART Flash RAM EEPROM Core(including interrupt and clock), etc., using the IC underlying library for standardized packaging
MCAL: Etherent CAN UART Flash RAM EEPROM Core(including interrupt and clock), etc
MODULE:Model_IO Retrieval Chains Memory Agents Callbacks
APP: An application is composed of an application chain and its corresponding modules
There are Plugins in each layer of the architecture, which can be third-party libraries or custom modules
Implementation of chain:
Context memory management:
LAN LLM access:
Local large model supporting the server program
Implementation of cloud multi-model protocol:
chatgpt
chatgml
文心一言
通义千问
Hardware support:
stm32f103+air742UG->4G
esp32-c3->wifi
Test Project
Raspberry RP2040 ->RJ45
Vector databases support:
Milvus(remote storage matching)
/TODO/
/TODO/
- Fork the repository
- Create Feat_xxx branch
- Commit your code
- Create Pull Request