This release has several stability improvements and features to enable AI teams to do more with MarkovML.

SDK Enhancements

In this release, we have several enhancements to Markov SDK. Please update to the latest version (1.4.5) of Markov SDK. The instructions to update are here

🚧

Update your SDK

Please update your SDK to v1.4.5

Dataset Upload Using SDK

With MarkovML, users can easily upload and register datasets using our SDK with just a few lines of code. We support uploading files and data frames and registering datasets stored on cloud storage such as Amazon S3. Your datasets will be uploaded to registered storage if you are an enterprise customer with Hybrid deployment.

Improved Dataset Interface

The Dataset interface has been improved for usability. You can now fetch the dataset details, including loading the details view of the dataset in the browser using Markov SDK. The code samples have been updated in the documentation here

Snippets Enhancements

Our collaborative reporting and documentation app has been enhanced to support Model Apps. With this change, you can share your Models as applications with people within your organization or publically using a shareable link.

Baseline Model Apps Enhancements

MarkovML's Baseline Model apps allow users to interact with and make inferences on labeled datasets. The experience has been enhanced to run inference on files and download results, which can also be reviewed within the platform.

Embedding View Enhancements

The newly introduced embedding labeling view allows users to choose specific points of interest from different areas of the space and label or relabel them within the platform. This feature is beneficial for visualizing the entire dataset and creating a representative sample for other purposes, such as training a few-shot classifier on an LLM.

UX Enhancements

We have revamped our listing view all across the product to improve aesthetics. In addition, we pre-emptively check for errors in the dataset during the registration process that might prevent this dataset from being analyzed by the platform. This saves time and improves the experience.

Stability & Bug Fixes

Several bug fixes across the platform have been addressed.