Installing Ollama Step by Step to run DeepSeek-R1 Offline

Tin Tin — Fri, 31 Jan 2025 18:01:29 +0000

What is Ollama?

Ollama is a tool that lets you run large language models (LLMs) like Llama 2, Mistral, or GPT-3 directly on your computer. Think of it as a "local ChatGPT" that doesn’t require an internet connection or cloud services. You can chat with AI, test ideas, or build apps without sharing data with third parties.

Why Run Ollama in WSL (Windows Subsystem for Linux)?

Most AI/ML tools (including Ollama) are optimized for Linux. Here’s why WSL is ideal for Ollama on Windows:

Native Linux Compatibility
- Ollama works best with Linux libraries and dependencies. WSL lets you run Linux commands and tools seamlessly on Windows, avoiding compatibility headaches.
Better Performance
- WSL 2 (the latest version) integrates tightly with Windows while offering near-native Linux speeds. This is crucial for running resource-heavy LLMs smoothly.
GPU Support
- Modern LLMs work faster with GPUs. WSL 2 supports GPU passthrough (e.g., NVIDIA CUDA), letting Ollama leverage your computer’s graphics card for faster responses.
Easy Setup
- Installing Ollama in WSL is as simple as running a Linux command. No complex workarounds for Windows-specific issues.
Keep Your Windows Environment Clean
- Isolate Ollama’s dependencies (like Python packages) within WSL. No clutter in your main Windows system!

Installing Ollama (The Engine)

💡Ollama is the foundational component for running AI models.

Windows Users (WSL)

We'll leverage WSL to create a Linux environment within Windows.

Open a "Terminal" or "Command Prompt" and run it as administrator. This is important for certain commands to function correctly.

Type wsl --install and press Enter. This command will install the latest version of WSL with the Ubuntu distribution of Linux.

After WSL is installed, you will need to restart your computer.
When you log back in, a new terminal window will open and continue the installation. You will be prompted to create a Linux username and password. These are different from your Windows username and password.

Your terminal is now running inside the Linux system.

Upgrade your WSL environment to newest versions

Type the following commands (one at a time, pressing Enter after each):
- sudo apt update (This command refreshes the package lists (also known as the "package cache") from the software repositories that your system is configured to use. Think of it as updating your grocery store's price list, not buying the items themselves). You will be asked for your password to run the sudo (root role).

- `sudo apt upgrade -y` (This command upgrades all upgradable packages on your system to their newest versions, based on the information obtained by`apt update`. The `-y` automatically confirms any updates)

Install Ollama from WSL terminal

curl -fsSL [https://ollama.com/install.sh](https://ollama.com/install.sh) | sh for installing Ollama.

After Installation (All Users):

The Llama API service will be running in the background and is configured to listen on port 11434. This is the port which will allow other programs to access it. You can check by open your Browser and navigate to http://localhost:11434

You can download different AI models using the command line. For example, to get the Deepseek-R1 model, you would type ollama pull deepseek-r1:8b. The downloaded model will be stored locally on your computer.

You can run the Deepseek-R1 model using the command line, you should type ollama run deepseek-r1:8b.

Mac and Linux Users

Download the correct version of Llama for your operating system from llama.ai and follow the installation instructions provided on the website.

Kusto Query Language (KQL)

Tin Tin — Thu, 30 Jan 2025 15:20:30 +0000

Why this?

This post marks my first major attempt at daily learning, summarizing, and posting to a public space what I've consumed, so I can track my progress day by day.

KQL came into my life in a rather jarring way. I saw a colleague named Kome at work doing fantastic things with something he called ADX, and I had absolutely zero knowledge of it. I had to overcome my embarrassment by starting to learn KQL and ADX in a very serious manner.

It's been one year since then, but what I've consumed has been so fragmented, and it's really annoyed me. So, I've decided to re-learn it from the beginning with a clear and comprehensive structure.

What is Kusto Query Language?
Kusto Query Language (KQL) is a powerful query language used to explore, analyze, and visualize data stored in Azure Data Explorer, Application Insights, Log Analytics, and other services that support the Kusto engine. It’s optimized for querying large datasets with minimal latency.

KQL allows you to:

Extract insights from structured, semi-structured, and unstructured data.
Perform complex aggregations, filtering, and transformations.
Visualize data using charts, tables, and time-series graphs.

Why Learn KQL?

Efficiency : KQL is designed for high-performance querying on large datasets.
Integration : It integrates seamlessly with Azure services like Azure Monitor, Azure Security Center, and more.
Ease of Use : Its syntax is simple and SQL-like, making it accessible for those familiar with SQL or other query languages.

2. Comparing KQL with Other Query Languages

Feature	KQL	SQL	Power Query (M)	Python (Pandas)
Primary Use Case	Analyzing logs, telemetry, and big data	Relational databases	Data transformation and ETL	General-purpose programming and analysis
Syntax	Declarative, pipeline-based	Declarative, set-based	Functional, step-by-step	Procedural, code-based
Performance	Optimized for large-scale analytics	Optimized for transactional queries	Moderate performance for small-to-medium datasets	Flexible but slower for very large datasets
Learning Curve	Easy for SQL users	Moderate	Steeper due to functional paradigm	Steepest due to general-purpose nature
Visualization	Built-in visualization tools (e.g., Azure)	Requires external tools (e.g., Power BI)	Integrated with Power BI	Requires libraries like Matplotlib/Seaborn

3. Setting Up Your Environment

To practice KQL, you’ll need access to a service that supports it. Here are some options:

Azure Data Explorer (ADX) : A fully managed service for real-time analytics.
- Create an ADX cluster and database in Azure Portal.
- Use the Kusto Explorer desktop app or the Azure Portal web interface.
Log Analytics : Part of Azure Monitor.
- Access via the Azure Portal under "Logs."
Application Insights : For monitoring application performance.
- Access via the Azure Portal under "Logs."
Free Sandbox : Microsoft provides a free KQL sandbox for learning purposes.

DEV Community: Tin Tin