watsonx Code Assistant

IBM watsonx Code Assistant for Red Hat Ansible Lightspeed December 2024 Release Introduces Updated Model for Faster, More Efficient Ansible Code Generation

The latest release of IBM watsonx Code Assistant for Red Hat Ansible Lightspeed introduces an updated cloud model to enhance the developer experience and optimize key features, such as single-task and multi-task prompting, playbook generation, and playbook explanation. Here’s a detailed breakdown of what’s new and how these changes affect your development workflows:

Updated 20B Ansible Model

IBM’s generative AI Large Language Foundation Models (LLM) are open-source, enterprise-level language models trained on an extensive data set that has been subjected to intensive pre-processing and careful analysis. With this release, IBM watsonx Code Assistant for Red Hat Ansible Lightspeed is updating its product capabilities to the 20B model to deliver significant improvements to your coding experience. Built on the open-source IBM 20B Granite model, IBM Research has further enhanced it by training it with Ansible-specific datasets and fine-tuning it for the WCA Ansible product, resulting in a more refined and efficient model.

Enhanced Feature Functionality

With the 20B model, users can expect notable improvements with single-task and multi-task generations. The model's larger size and expanded training data set enhance its foundational knowledge of both natural language instructions and coding concepts. The new model also features an expanded, 8k token, context window, which should be beneficial when generating tasks within larger playbooks. These optimizations are designed to streamline daily coding tasks, saving you time and reducing friction, so you can concentrate on what matters most – writing code

Technical Updates

Previously, single-task and multi-task features were processed by the 3B model. With this release, the 20B model takes over the processing of these features, as well as Playbook generation and Playbook explanation. For customers on the Essentials plan, all watsonx Code Assistant for Red Hat Ansible Lightspeed features will now be powered by the updated 20B model. Additionally, on-premises customers will benefit from a simplified installation process, requiring only a single model installation to access the entire feature set.

Customization Options with the 3B Model

For users who wish to customize and train the IBM Ansible base model with their own datasets, they will be working with the 3B model. Once customized, the single-task and multi-task features will be processed by the 3B model. Please note that choosing to customize will require the installation of an additional GPU to support the 3B tuned model. This will enable users to leverage their tailored model for specific use cases, while still benefiting from the core features of the watsonx Code Assistant for Red Hat Ansible Lightspeed.

Seamless Model Updates to Existing Users

Our Essentials and Standard plans will be updated to this new model automatically, meaning no user intervention needed to re-install or upgrade models on SaaS. On premises availability will require reinstallation of the new CloudPak for Data (CPD) 5.1 and will offer a decoupled system with allows you to use tuning optionally with less GPUs.

Deployment Timeline and What’s Available

This new model and feature optimizations have been rolled out across all major deployment options:

• SaaS (as of December 5th, 2024)
• On-premises (as of December 11th, 2024)

What’s next?

New users can sign up for the free 90-day trial or explore the interactive demo to see IBM watsonx Code Assistant for Red Hat Ansible Lightspeed in action. Already using the new features? Leave a review and receive a $25 gift card! Let us know how the updates are working for you.