AI Fabric 20.10 Release

AI Fabric released cloud in June, on-premises single node in August, air-gapped and RedHat Enterprise Linux support in September. In the 20.10 long term support (LTS) release, we’ve enhanced our on-premises offerings with multi-node support – Microsoft Azure AKS, Docker Enterprise Edition (Docker EE).

The multi-node version will let you run lots of ML Skills and training pipelines in an environment with high-availability and elastic processing power. Both Microsoft Azure AKS and Docker EE are very popular managed Kubernetes platforms. We will be adding more most-used Kubernetes platforms in the coming releases.

We’ve also streamlined AI Fabric on-premises installation with less reliance on Orchestrator, fewer manual steps besides provisioning machines, and more robust installation.

Besides DU models (Purchase Order, Utility Bills and others), we’ve added more starter models, which include Semantic Similarity and new Tabular Data models. Semantic Similarity model shows which sentences or words correlate with each other. A common use case is to analyze recurring trends in product reviews. We also released Tabular Data classification and regression models. These models work with data in CSV files.

Highlight of this release:

  • On-premises multi-node: Microsoft Azure AKS, Docker EE
  • Streamlined on-premises installation
  • More models: Semantic Similarity, Tabular Data models, Intelligent Keyword Classifier, Intelligent Form Extractor, and Handwriting OCR
  • Diagnostic tools: run health checks for connected installation on your clusters to show how well things are working and identify issues.
  • Integration with Data Manager (preview)
  • Public endpoints for ML skills: at the click of a button, a URL will be generated. The URL allows customers to use an ML skill without having to use robots, or when their robots are not connected to the same Orchestrator tenant.

New content/events


No proper information about “Semantic Similarity” on the documentation portal.