Virtualitics AI Platform
Version 1.23 (2024.4)
-
Predict Product Rename: Virtualitics Predict has been officially renamed to the Virtualitics AI Platform as the first step in a larger effort to unify the Virtualitics product suite.
-
Flow to App Rename: Applications built with the Virtualitics SDK will be referred to as “Apps” and each run of an App (stored in the App History table) will be considered a “flow”.
-
New Task Execution Framework: Replaced Celery with a new system for executing and tracking Flows, Steps, and Events, offering improved task management and performance across the platform.
-
Virtual Environment Support: Introduced a unified virtual environment to enhance dependency isolation and management when executing tasks, with support for multiple environments coming in a future release.
-
Deprecation of Self Service Flows: Self Service Flows written using RestrictedPython are no longer supported on the platform and the "Create Flow" file uploader is removed from the home page. We now support Custom Apps, which are deployed to our platform using the Virtualitics CLI tool.
Virtualitics Developer Experience
Virtualitics SDK
-
Auto Complete for App Development: Pip install the virtualitics-sdk stubs Python package to begin writing Custom Apps with improved autocomplete, type checking, documentation tooltips, and parameter lists for better developer guidance.
-
App Management through the CLI: Pip install the virtualitics-cli Python package to build, package, and deploy Custom Apps from the command line in a local development environment.
-
build
: Builds a VAIP App Python wheel file. -
config
: Used to create a configuration file. -
deploy
: Deploys the VAIP App to a VAIP Instance. -
destroy
: Deletes a VAIP module, and all the apps of that module. -
init
: Initializes a VAIP app structure. -
use-context
: Used to set the context referenced in the configuration file.
-
-
Concurrency Support: Users can implement their own concurrency (multiprocessing, threading, and asynchronous programming) within Apps.
-
Rich Formatting in Table Cells: Data grid tables now support Markdown, allowing users to improve readability with rich text formatting (bold, italics), add links, images, code snippets, and math equations, and apply alignment options (left, right, center) for a more dynamic and customizable data presentation.
-
Expanded Plotly Plot Options: Leverage an expanded array of Plotly plot options within your apps, such as choropleth plots, heat maps, candlestick plots, and more.
-
Deep Learning Libraries Available: Virtualitics app writers now have access to popular deep learning libraries such as PyTorch, TensorFlow, Transformers, and Keras to perform deep learning tasks within Apps.
-
Polars Library Available: Virtualitics app writers can now leverage the Polars DataFrame library to manipulate structured data within Apps.
-
Stability and Bug Fixes:
-
Breaking Changes:
-
PlotOrientation
removed from the SDK, which affects plot creation functions likecreate_bar_plot
-
Card.content
ratio system has been adjusted to the correct nesting structure (content =List
[Union
[Element
,Row
]], Row =List
[Union
[Element
,Column
]]) -
as_columns
is removed as an argument ofCard
, soColumn
s can now only be implemented within aRow
-
PageType
removed from the SDK, which affects the instantiation of aPage
-
Page
no longer accepts themetadata
parameter in its constructor -
Any apps with app-specific documentation will need to update their
docs_index.json
to replacedeployed_flow_id
withapp_id
-
-
Backwards Compatible Changes:
-
Flow
class changed toApp
class
-
-
Virtualitics Explore
Version 1.23 (2024.4)
Explore
- Improved support for datasets in cloud storage: Datasets can be loaded from the Virtualitics cloud storage system into an active project, and can be saved to that system from within a project.
- Home Screen Updates: Revised home screen to work better with the above improvements for working with datasets in the cloud storage system (Persistence).
- Pathfinder History: Project files (.vip) now contain Pathfinder results so that they can be shared and referenced after you save a project.
- Worker system: Explore can now support more concurrent users thanks to this new worker system.
- Dataset and project size limits: These are configurable per deployment, with defaults of 1.5M rows and 2k columns per dataset, and 10M rows, 10k columns and 20 datasets per project.
- Removed "0 - " and "1 - " from Anomaly Detection output column.
- General stability and bug fixes.
Explore Desktop
- Deprecation of VR: VR is no longer supported.
- Deprecation of SVO: Shared Virtual Office (SVO) for real-time collaboration is no longer supported.
- Removed "0 - " and "1 - " from Anomaly Detection output column.
- General stability and bug fixes.