How to use DeepSeek 🐋 | Beginner Friendly Tutorial | Free Course

Introduction

This tutorial will guide you through using DeepSeek, a powerful open-source AI model that rivals OpenAI's latest offerings. By the end of this guide, you'll be able to set up DeepSeek for free, test its capabilities, and determine if it's the right fit for your projects.

Step 1: Use DeepSeek On Your Browser

• Visit the DeepSeek web app.
• Familiarize yourself with the interface and available features.
• Test simple queries to understand how DeepSeek processes requests.

Practical Tip: Start with basic questions to gauge response accuracy before diving into complex queries.

Step 2: Use DeepSeek Privately

• Download and install the Ollama software to run DeepSeek locally on your machine.
• Follow the installation instructions provided on the Ollama website.
• Once installed, initiate DeepSeek by running the following command in your terminal:

  ollama run deepseek
  

• This allows you to use DeepSeek without an internet connection, enhancing privacy and control.

Common Pitfall: Ensure your system meets the necessary requirements for running Ollama to avoid installation issues.

Step 3: DeepSeek in the Terminal

• Open your terminal.
• Use the command line to interact with DeepSeek directly.
• You can input queries and receive responses in real-time.

Example Command:

echo "Your query here" | ollama run deepseek

Practical Tip: Experiment with different queries to see how DeepSeek handles various topics.

Step 4: Use Chatbox with DeepSeek

• Access the Chatbox feature for a more user-friendly experience.
• This interface allows for back-and-forth conversations, mimicking a chat experience.
• Input queries and explore DeepSeek's responses interactively.

Common Pitfall: Ensure you understand the context of your queries for better responses.

Step 5: Temperature Test

• Experiment with different temperature settings to modify response creativity.
• Lower temperatures (e.g., 0.2) yield more deterministic responses, while higher temperatures (e.g., 0.8) provide more creative outputs.

How to Set Temperature:

ollama run deepseek --temperature 0.7

Practical Tip: Adjust the temperature based on your project's requirements—higher for creative tasks, lower for factual tasks.

Token Efficiency Showdown

• Compare the token efficiency between DeepSeek and ChatGPT.
• Analyze the output quality relative to the number of tokens used.
• This comparison can help you determine which model suits your needs better.

Delete Your Resources

• If you're running DeepSeek locally and wish to free up resources, you can delete unused models or installations.
• Use the following command to clean up:

ollama delete model_name

• Replace model_name with the specific model you want to remove.

Conclusion

In this tutorial, you've learned how to set up and utilize DeepSeek effectively, both online and offline. You've explored its features, tested its performance against ChatGPT, and learned how to adjust settings for optimal use.

Next steps include experimenting further with DeepSeek, joining the NextWork community for additional support, and considering how DeepSeek can enhance your AI projects.