Use the Custom Models feature to consolidate pre-processing steps, trained models, and post-processing steps for a specific task into a single package.

You can either use an existing trained model from MarkovML or create a new one to build your custom model package. Once you have created your custom model package, you can register it with the Markov model registry. This will allow you to access your custom model from the MarkovML UI platform for future deployment or tracking purposes.

No more time-consuming task of redoing pre-processing and post-processing steps every time you need to use them. With MarkovML, you can effortlessly store, deploy, share, and collaborate with your team.

All you have to do is build your Model Inference Pipeline.

How it Works

The process begins with you defining your inference model name and building your inference stages. MarkovML provides operators to specify the type of action being performed during each stage.

MarkovML executes tasks using your inference stages in the given order and packages them into one bundle called custom models.

You can then register your custom model to Markov SDK and view it on the MarkovML UI platform.