Skip to main content

Daily Workflow

Your typical routine for working with DS01 containers.

The Core Pattern​

# Start your work session
project launch my-project --open

# ... do your work in the container ...

# End your session
exit
container retire my-project

That's it. Two commands to start, two to finish.

Starting Your Session​

project launch my-project --open

Smart: Checks if image exists, offers to build if needed, then launches.

Option 2: container deploy (Direct)​

container deploy my-project --open

Direct: Requires image to already exist.

Difference: Use project launch if you're not sure whether the image exists.

Working in Your Container​

Once inside your container:

# You're at /workspace - your project files
cd /workspace

# Run Python scripts
python train.py

# Start Jupyter
jlab

# Use Git normally
git status
git commit -m "progress"

Your files are at /workspace - this maps to ~/workspace/my-project/ on the host.

Ending Your Session​

Exit the container​

exit
# or press Ctrl+D

The container keeps running in the background.

Choose what to do next​

Keep running (default): Container stays active, GPU allocated

# Reconnect anytime
container-attach my-project

Retire (recommended when done): Frees GPU for others

container retire my-project

Quick Reference​

ActionCommand
Start workingproject launch my-project --open
Exit containerexit or Ctrl+D
Reconnectcontainer-attach my-project
Done for the daycontainer retire my-project
Check running containerscontainer-list
Check resource usagecontainer-stats

Tips​

Save work frequently​

Files in /workspace persist. Container changes (pip installs) don't.

Add packages permanently​

image-update my-project --add "wandb optuna"
container retire my-project
project launch my-project --open

Background mode​

project launch my-project --background
# Access via VS Code Remote or container-attach

Next Steps​