Rasa, the customizable open source machine learning framework to automate text and voice-based AI assistants, has released version 2.0 with significant improvements to dialogue management, training data format, and interactive documentation. In addition, the latest release reduces the learning curve to get started while expanding configuration options for advanced users.
Rule Policy
Rasa Open Source 2.0 simplifies dialogue policy configuration, draws a clearer distinction between policies that use rules and those that use machine learning, and makes it easier to enforce business logic. Previously, rule-based logic in Rasa Open Source was controlled by a combination of three or more dialogue policies. The new RulePolicy allows users to implement forms, map actions to intents, and specify fallback logic, using a single policy.
Form Updates
The forms implementation has been moved from the Rasa SDK into the Rasa Open Source library. This makes the Python Rasa SDK more lightweight, moves essential functionality into the main library, and makes it easier for developers to implement action servers in new programming languages.
YAML Training Data
Rasa Open Source now supports YAML for training data. This offers three advantages: the ability to support custom metadata in training examples, like the user who added it and timestamp; the ability to divide long files into smaller, more modular ones; and the ability to support rich media in the response selector.
Suggested Config
Rasa Open Source now simplifies installation by recommending a default NLU pipeline when you initialize a new project. The suggested configuration can be overridden by advanced users to further customize the pipeline.
Retrieval Intents
Retrieval intents were introduced in Rasa Open Source 1.3 as an experimental feature, to easily handle single-turn interactions like chitchat and FAQs. Rasa 2.0 provides full support for retrieval intents, including rich media in responses, like images and buttons, as well as full Rasa X support.
Documentation and Rasa Playground
Alongside the 2.0 release, the Rasa documentation has received a new theme and updated information architecture. The new layout was designed with both beginning and returning users in mind, to make relevant articles more discoverable.
The new documentation also includes the addition of the Rasa Playground. The Playground allows users to prototype new projects directly in the browser, without installing Rasa Open Source first. Once a prototype has been built, the project files can be downloaded for continued development.
Migrating to Rasa Open Source 2.0
Rasa has also released a guide to migrating from 1.10 to 2.0 for users upgrading existing assistants. The guide includes instructions for converting training data files from markdown to YAML, as well as how to update your dialogue policies and training data to use the new RulePolicy.
In addition, Rasa Open Source 2.0 is fully compatible with Rasa X, a free, closed source UI tool that helps developers improve AI assistants. Rasa X now includes support for retrieval intents, which now appear alongside regular intents in the UI. Rasa X users can also view and edit multimedia responses.
Rasa Open Source 2.0 can be installed via pip and is available on GitHub.