Skip to main content

Using the Help System

DS01 commands have built-in help to guide you from beginner to expert.

Four Help Modes​

Every command supports four help modes - two for quick reference, two for learning:

FlagTypeWhen to UseWhat You Get
--help or -hQuick ReferenceKnow the command, need syntaxUsage, main options, examples
--infoFull ReferenceNeed all optionsComplete documentation, all flags
--conceptsLearn FirstNew to this conceptExplains what something is before running
--guidedLearn by DoingFirst time using commandStep-by-step with explanations

Quick Reference: --help​

When: You know what the command does, just need the syntax.

Example:

container-deploy --help

Shows:

DS01 Container Deploy
L3 Orchestrator - Creates and starts containers in one command

Usage:
  container-deploy [name] [options]

Options:
  --guided          Show detailed explanations
  --background      Start without attaching
  --open            Start and open terminal
  -h, --help        Show this help

Examples:
  container-deploy                  # Interactive wizard
  container-deploy my-project       # Quick deploy
  container-deploy my-project --open

Fast, concise, straight to the point.

Full Reference: --info​

When: You need to see ALL options and detailed examples.

Example:

container-deploy --info

Shows:

  • Complete option list
  • All subcommands
  • Multiple examples
  • Advanced use cases
  • Related commands

Like a man page, but friendlier.

Learn the Concept: --concepts​

When: You're new to containers, images, or DS01 concepts.

Example:

image-create --concepts

Shows:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
Understanding Docker Images
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

What is a Docker Image?

An image is like a recipe for creating containers. It defines:
  • What software is installed (PyTorch, pandas, etc.)
  • What system libraries are included
  • How the environment is configured

Think of it as a snapshot of a configured system.

Container vs Image:
  Image     →  Recipe/Blueprint (stored, sharable)
  Container →  Running instance (temporary, for work)

You can create many containers from one image.

Why Images Matter:

1. Reproducibility - Same environment every time
2. Shareability - Give Dockerfile to colleagues
3. Version Control - Track changes in git
4. Efficiency - Rebuild containers instantly

[Press Enter to continue or Ctrl+C to exit]

Read this BEFORE running the command - builds understanding first.

Learn by Doing: --guided​

When: First time using a command or want explanations during execution.

Example:

image-create --guided

What happens:

1. Explains Before Each Choice​

━━━ Framework Selection ━━━

PyTorch, TensorFlow, and JAX are deep learning frameworks.

PyTorch:
  • Most popular for research
  • Dynamic computation graphs
  • Great for experimentation

TensorFlow:
  • Industry standard for production
  • Static computation graphs
  • Extensive tooling

JAX:
  • Cutting-edge research
  • Fast on GPUs/TPUs
  • Functional programming style

Which framework do you want?
  1) PyTorch 2.8.0 (recommended for research)
  2) TensorFlow 2.16.1
  3) JAX 0.4.23

[Press Enter to see options]

2. Pauses for You to Read​

Press Enter to continue...

Gives you time to understand before proceeding.

3. Shows What's Happening​

Building Docker image...
────────────────────────────────────────────────
  [Step 1/4] Downloading PyTorch base image...
  [Step 2/4] Installing system packages...
  [Step 3/4] Installing Python packages...
  [Step 4/4] Configuring environment...
────────────────────────────────────────────────

4. Explains What Just Happened​

━━━ What Just Happened? ━━━

Your Docker image was built with:
  • PyTorch 2.8.0 with CUDA 12.4
  • pandas, numpy, matplotlib
  • Jupyter Lab
  • Your custom packages

The image is stored as: ds01-username/my-project:latest

You can create containers from this image anytime with:
  project launch my-project

Best for first-time users - learn while doing.

Interactive Mode (No Arguments)​

When: You're not sure what arguments to provide.

Just run the command without anything:

# These all start interactive wizards
project init
container deploy
image create

What happens:

  • Command asks questions
  • Presents menus
  • Guides you through options
  • No need to memorise flags

Example:

container-deploy

Shows:

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
DS01 Container Deploy
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Select a project to deploy:

  1) my-thesis          (PyTorch 2.8.0)
  2) research-2024      (TensorFlow 2.16.1)
  3) experiments        (JAX 0.4.23)

Choice [1-3]:

Friendly, menu-driven, no flags needed.

Quick Command Reference: commands​

See everything available:

commands

Shows:

  • All DS01 commands
  • Organised by category
  • Brief descriptions
  • Common workflows

Like a cheat sheet - quick lookup of what's available.

Combining Help Modes​

Pattern 1: Beginner Learning​

# 1. Learn concept first
image-create --concepts

# 2. Run with guidance
image-create --guided

# 3. Second run: interactive
image-create

# 4. Later: Quick reference
image-create --help

# 5. Finally: experienced, run non-interactively as CLI
image-create my-project -f pytorch -t nlp

Help for Subcommands​

Dispatchers have help too:

# See all container commands
container help

# Help for specific subcommand
container deploy --help
container retire --help

Also works:

# Hyphenated form
container-deploy --help

Both formats are equivalent.