We're excited to announce the arrival of a new and improved Gradient Notebooks experience -- now with Gradient Datasets, native support for interactive widgets, improved cell, file, and kernel management experiences, and much more!
Highlights are below but be sure to read the blogpost for the most detailed explanation of this release.
Introducing Gradient Datasets announcement
We're pleased to announce the arrival of Gradient Datasets! Datasets make it easy to generate portable datasets to use across Gradient teams and resources.
You can now create and mount datasets for easy use within a notebook and take advantage of a number of public datasets made available by the Paperspace team.
Check out the blogpost for a full list of public datasets.
Support for interactive widgets improvement
Gradient Notebooks now provides first-class support for ipywidgets! This includes sliders, checkboxes, multiselects, TensorFlow and PyTorch dataloaders, and more!
The full list of supported widgets is available here.
Cell management improvements improvement
We've brought over to notebooks a number of cell management operations from JupyterLab such as insert, join, split, and more!
We'll be continuing to add cell management capabilities to the new IDE over time.
File management improvements improvement
In addition to cell management improvements, we've also made it easier to manage and manipulate files within the notebook file browser.
The file manager now behaves as expected when dragging and dropping files and folders.
Kernel management controls improvement
We've improved the controls for starting and stopping individual kernels from within a notebook.
It's now easy to assign a notebook file to a particular kernel and to restart and stop individual kernels!
Bonus for Pro/Growth users: terminal updates! improvement
For users on the Pro or Growth plan, we've enabled split-screen terminals!
Now it's possible to work in a terminal without leaving a notebook file!
More improvements improvement
- We improved resource allocation and decreased notebook pending timeouts for notebooks which means higher availability of notebook machines and fewer stalled notebook starts
- We improved the refresh rate of notebook logs and improved notebook metrics to display more useful information
- We updated the two most popular runtime tiles in Gradient Notebooks: PyTorch and TensorFlow! The latest distribution takes advantage ofPyTorch 1.11 and TensorFlow 2.7.0.
- We fixed an issue that sometimes caused users to be signed out of the Paperspace console when swapping between tabs or sessions
- We fixed an issue that sometimes caused users with multiple teams to view incorrect resource data
- We fixed an issue that sometimes caused deployment items to expire and be deleted on teams with a large number workflows and deployments read the blogpost