Lecture Notes on DeepSeek R1 and Local Model Running

Introduction to DeepSeek R1

Recently caused significant discussion and market reactions.
Recognized as the best free and open-source AI model available.
Privacy Concerns: Using the model on its official website stores user data on servers in China, which have a loose privacy policy.

Advantages of running DeepSeek R1 locally:
- Enhanced privacy (no data leaves your computer).
- Easier access to powerful computing capabilities.

Download LM Studio:
- Visit the LM Studio homepage and download the version for your computer.
Open LM Studio:
- Navigate to the "Discover" tab.
- Type "DeepSeek" in the search bar.
Choose Model Specifications:
- Various options for model sizes based on hardware capability.
- Full DeepSeek R1: Contains 671 billion parameters; requires high-end hardware.
- Example Model: 8 billion parameters model recommended for average computers.
- More powerful configurations can be achieved with additional hardware (e.g., RTX 5090).
Select and Download Model:
- Choose a model based on quantization levels (e.g., Q4KM1).
Begin Prompting:
- Ensure the correct model is selected in LM Studio.
- Enter prompts and enjoy full local functionality.

Using DeepSeek via deepseek.com:
- Process involves logging in, using prompts, and observing reasoning.
- Example prompt: "If a plant that doubles its size every day covers a lake in 30 days, how much time will it take for two plants to cover half the lake?"
- Response time: 92 seconds, showcasing human-like reasoning.

Running DeepSeek locally is straightforward and maintains user privacy.
The model demonstrates capability across various tasks beyond simple prompts, including creativity in poetry and songwriting.
For deeper learning on AI, courses available on Futurpedia.
- Access: Free courses to enhance skills in AI integration into life and career.