An intelligent agent that leverages the Hugging Face quantized model hugging-quants/llama-3.2-3b-instruct hosted on LM Studio to answer user queries. For tasks where the model cannot access real-time information (like the current time), it automatically invokes an internal tool to provide accurate responses.
- Chat with a local LLaMA 3.2B instruct model using LM Studio
- Custom tool integration for real-time data (e.g., current time)
- Asynchronous agent execution for fast responses
- Fully modular: easily add more tools in the future
device-agent-ai/
├─ main.py # Entry point for the agent
├─ src/
│ ├─ agent.py # Agent logic and orchestration
│ ├─ model.py # LM Studio model interface
│ └─ tools.py # Internal tools (e.g., current time)
├─ requirements.txt # Python dependencies
└─ README.md
