DeepSeek: Your Local AI Powerhouse with Ollama

The world of Large Language Models (LLMs) is rapidly changing, and running them locally is becoming increasingly appealing. DeepSeek, a powerful open-source LLM, combined with Ollama, makes this process surprisingly simple. This guide will walk you through setting up DeepSeek on your own machine.

DeepSeek: A Powerful LLM

DeepSeek stands out for its strong coding abilities and general language understanding. It's open-source, allowing for transparency and community contributions, and importantly, it can be run locally, it’s also the first AI that took an AI’s Job xD(ChatGPT).

Ollama: Simplifying Local LLM Use

Ollama acts as a container and runtime for LLMs, making local deployment straightforward. It handles dependencies and configurations, allowing you to focus on using the model, Docker can also be used :)

Benchmark below:

Getting Started:

1. Install Ollama:

   • macOS: Open your terminal and run: brew install ollama

   • Linux: Open your terminal and run: curl -fsSL https://ollama.ai/install.sh | sh

   • Windows: Follow the installation instructions from the official Ollama website.

2. Pull DeepSeek:

   • Open your terminal and run: ollama pull deepseek-r1:1.5b

3. Run DeepSeek:

   • In your terminal, run: ollama run deepseek-r1:1.5b

for example:

4. Interact in the Terminal:

   • Ask questions or provide instructions. Example: "Write a python function that reverses a string."

5. Stop DeepSeek:

   • Press Ctrl+D (or Cmd+D on macOS).

6. Using DeepSeek in a Script:

   • You can integrate DeepSeek into your Python scripts: Python

       import ollama
       response = ollama.chat("deepseek-r1:1.5b", "Hello, how can you assist me?")
       print(response)
     

7. Using a Web UI:

   • For a better user experience, consider using Open WebUI: https://github.com/open-webui/open-webui

   • After installation, you can access the UI at http://localhost:11434/api/chat.

Important Considerations:

• Running LLMs locally requires sufficient hardware resources.

• Experimenting with different prompts will help you achieve better results.

DeepSeek and Ollama together provide a powerful and accessible way to use LLMs locally.

Models like deepseek will keep coming in the market, it’s us (coders) who have to improve :D, do share your opinion on this statement in the comments!!

Till then, explore, take care, learn and enjoy 🔥

Extra links:

Discord: https://discord.gg/hackunited

Thanks a lot for checking this out, we hope that you have fun! :D