Notebooks How-Tos
Notebooks are a web-based Jupyter IDE with shared persistent storage for long-term development and inter-notebook collaboration, backed by accelerated compute.
Getting Started
Create a notebook using Paperspace’s CPU- and GPU-backed machines.
Swap and see the kernel state for machines attached to notebooks.
Fork a notebook to create a clone of a public notebook into your workspace or to duplicate a notebook already in your workspace.
Collaboration
Share access to Notebooks to let other users view and run your notebook.
Run on Gradient is the fastest way to share Gradient Notebooks on Paperspace.
Manage Data
Create datasets using the Paperspace console and command line.
Mount existing team datasets, public datasets, and create new team datasets to explore data and training models.
Upload and download files from the file manager, transfer files from Google Drive to your notebook, and access shared persistent storage.
Mount public or private S3 buckets into notebooks to access data stored externally.
Manage Storage
Add DigitalOcean Spaces Object Storage or another public storage provider.
Manage storage in Linux-based environment.
Configuration
Access a terminal in Gradient Notebooks to use root access to the underlying machine.
Connect to remote Jupyter kernels through alternative IDEs.
How to check on Jupyter kernel states
Manage containers to ensure reproducible results when using Notebooks, Workflows, and Deployments.
Logs and Metrics
Access real time and post-workload system logs for Gradient Deployments.
Access CPU usage, RAM usage, and GPU performance metrics for Gradient Notebooks.
Use TensorBoard, a visualization toolkit from TensorFlow, within Gradient notebooks to visualize metrics and histograms, display images, and more.