Intro to Deep Learning with PyTorch

Register: https://www.zoom.com/

Zoom is a video conferencing software that allows for virtual meetings and webinars. It is essential for attending our online workshop sessions and provides the interactive features needed for effective learning.

Why Zoom?

• Breakout Rooms: Essential for our small-group exercises
• Screen Sharing: Share your screen to get help or demonstrate solutions
• Stable & Reliable: Handles large groups with consistent quality
• Recording: Sessions can be recorded for later review (where permitted)

Installation

Download and install the Zoom Desktop Client from the official website. We require the desktop client rather than the web version for full feature support.

Before Your First Session

1. Test Your Setup: Join a test meeting to check audio/video
2. Update Zoom: Make sure you have the latest version
3. Check Your Internet: Ensure you have a stable connection
4. Find a Quiet Space: Minimize background noise during sessions

Workshop Etiquette

• Keep your microphone muted when not speaking
• Use video when possible to help build community
• Use reactions (👍, ✋) to provide feedback
• Ask questions in chat or unmute to speak
• Be ready to join breakout rooms for exercises

Tips

• Familiarize yourself with screen sharing features before the workshop
• Keep your Zoom name consistent with your registration
• Use virtual backgrounds if needed for privacy
• Enable “dual monitor mode” if you have two screens

Register: https://pixi.sh/latest/installation/

Pixi is a modern package manager that simplifies the installation of scientific software. It’s built on top of the conda-forge ecosystem but is significantly faster and provides better dependency resolution.

Why Pixi?

• Fast: 5-10x faster than conda for most operations
• Reproducible: Uses lock files to ensure exact environment reproduction
• Task Runner: Built-in task management (like npm scripts)
• Modern Design: Clean CLI with better error messages
• Conda Compatible: Uses the conda-forge repository

Installation

Follow the installation instructions on the official Pixi website. The installer will set up Pixi and configure your PATH automatically.

Getting Started

Initialize a new project:

pixi init
pixi add python
pixi shell

Add packages:

pixi add numpy pandas matplotlib

Define and run tasks in pixi.toml:

[tasks]
dev = "python main.py"
test = "pytest tests/"

Run tasks:

pixi run dev
pixi run test

Advantages Over Conda

• Much faster package resolution and installation
• Lock files ensure reproducibility by default
• Better support for managing multiple projects
• Built-in task runner eliminates need for separate tools

Tips

• Use pixi shell to activate the environment
• Define common tasks in pixi.toml for easy project workflows
• Lock files (pixi.lock) should be committed to version control
• Use pixi global install for system-wide tools

Register: https://git-scm.com/downloads

Git is a version control system that tracks changes in source code. It allows multiple people to work on a project simultaneously and maintains a complete history of all changes.

Why Git?

• Distributed: Every developer has a complete copy of the project history
• Branching: Experiment with new features without affecting the main codebase
• Collaboration: Work with others seamlessly through platforms like GitHub
• Reproducibility: Track exactly which version of code produced which results

Installation

Download and install Git from the official website. Choose the appropriate installer for your operating system.

Windows

Use Git for Windows installer with recommended defaults.

macOS

Git comes pre-installed on most macOS systems. Update with Homebrew: brew install git

Linux

Install using your package manager: sudo apt-get install git (Ubuntu/Debian)

Configuration

After installation, configure your identity:

git config --global user.name "Your Name"
git config --global user.email "your.email@example.com"

Essential Commands

• git clone - Copy a repository to your local machine
• git add - Stage changes for commit
• git commit - Save changes with a message
• git push - Upload changes to remote repository
• git pull - Download changes from remote repository

Tips

• Use meaningful commit messages that explain why you made changes
• Commit frequently to create detailed checkpoints
• Create branches for new features or experiments
• Use .gitignore to exclude data files and generated content

Register: https://github.com/

GitHub is a web-based platform built around Git that provides hosting for software development and version control. It’s the world’s largest code hosting platform and essential for modern collaborative research.

Why GitHub?

• Collaboration: Work with researchers worldwide on shared projects
• Visibility: Make your research code discoverable and citable
• Integration: Connect with CI/CD, documentation, and project management tools
• Community: Access to millions of open-source projects and libraries
• Free for Research: Unlimited public and private repositories

Getting Started

1. Create a free account at github.com
2. Set up Git on your local machine
3. Configure Git with your GitHub credentials
4. Create your first repository or clone an existing one

Essential Features

Repositories

• Host your code with full version history
• README files for documentation
• Issues for tracking bugs and features
• Pull requests for code review

Collaboration

• Fork projects to contribute
• Star repositories to bookmark them
• Follow researchers working in your field
• Use GitHub Pages for project websites

Tips for Researchers

• Include a LICENSE file to clarify how others can use your code
• Write a clear README explaining what your code does
• Create a CITATION.cff file for proper attribution
• Use releases to mark versions associated with publications
• Add topics to make your repository discoverable

Best Practices

• Commit often with meaningful messages
• Use branches for new features
• Write clear documentation
• Add a DOI through Zenodo integration for permanent archiving

Register: https://www.sciebo.de/

Sciebo is a cloud storage service for universities in North Rhine-Westphalia, Germany. It provides secure, GDPR-compliant storage for research data with large storage quotas.

Why Sciebo?

• Secure: Hosted in Germany with GDPR compliance
• Generous Storage: Large quotas for academic users
• University Integration: Uses your university credentials
• Collaboration: Share files and folders with colleagues
• Sync Across Devices: Desktop and mobile apps available

Features

• File synchronization across devices
• Sharing via links with password protection
• Collaborative document editing
• Version history for files
• Integration with university authentication

Getting Started

1. Access Sciebo through your university’s login
2. Install the desktop sync client (optional)
3. Create folders for organizing your research data
4. Use sharing features to collaborate with colleagues

Best Practices

• Organize files in clear folder structures
• Use descriptive file names with dates
• Set appropriate sharing permissions (read vs. edit)
• Regularly backup important data to multiple locations
• Be mindful of data sensitivity and compliance requirements

Tips

• Use selective sync to save local disk space
• Share folders instead of individual files for projects
• Use public links for sharing with external collaborators
• Check your storage quota regularly