Enhance AI Digital Assistant Capabilities with Dell APEX Cloud Platform for Red Hat OpenShift
Wed, 08 May 2024 17:37:41 -0000
|Read Time: 0 minutes
Boosting AI app development
It is no secret that AI applications and models tend to be hungry for storage capacity, whether that is to hold data for training, to store trained models, for additional data sets for Retrieval-Augmented Generation (RAG), or for incoming data to be processed by the system. Big models require lots of GPU memory and even bigger datasets on shared network-accessible storage systems for training.
With its continued support for Dell PowerFlex block storage, and now with support for S3 object storage using solutions such as Dell ObjectScale, the latest release of the Dell APEX Cloud Platform for OpenShift can unlock more flexible storage possibilities to help you do more with your data.
Also, with newly introduced support for NVIDIA L40s GPUs, we are making it a bit easier to keep up with model size inflation and providing more concurrency and scalability for applications. These enhancements make Dell APEX Cloud Platform for OpenShift an ideal foundation for developing and deploying AI applications.
The Dell validated design digital assistant gets a bigger brain
To showcase the new hardware and storage support, the Dell Validated Design for a digital assistant has been updated to use the Llama2-13b model stored on Dell object storage.
Our Dell Technologies Validated Designs are tested and proven configurations, designed from the start to dynamically fit needs based on specific use cases. These integrated solutions have been stringently tested and documented to help speed and simplify deployment of new solutions. By offering IT solutions with flexible design choices and guidance on choosing the right components, these Validated Designs can shorten deployment timelines - reducing, or in some cases eliminating, the time it takes to design, test, and integrate components.
The updated digital assistant design builds on the microservice architecture available with OpenShift and leverages the AI application building and serving capabilities of OpenShift AI, including the use of data science pipelines.
The pipelines can be created in Jupyter notebooks, exported, and then imported into OpenShift AI to enable repeatable and schedulable automated execution runs. In this case, we use it to poll a network file share, find newly added documentation, and update the embedding vector database with the new results. This keeps our results more relevant as information changes, because base models tend to lag behind the most current and up-to-date knowledge.
To show all the possibilities available to tweak an AI assistant, the UI has been updated with some new bells and whistles. It now allows you to select different RAG document stores to keep different categories of ingested documents separate. (Think of wanting to make sure that the engineering department doesn’t have access to the finance department’s documents, and vice versa.) You can change the model itself and tune additional parameters for creativity or accuracy.
AI audio transcription: Speak and spell with NVIDIA Riva speech services
AI audio transcription and translation is another solution with business value potential across many industries. Here it is easy to see how NVIDIA Riva automatic speech recognition and natural language processing, running on Dell APEX Cloud Platform for OpenShift with L40S GPUs, can keep everyone on the same page and solve problems that cross language barriers.
The Dell Reference Design for NVIDIA Riva speech services on OpenShift AI shows not only how OpenShift AI can be used to test and become familiar with AI technology, but also how after application development is complete, OpenShift builder can take your code and quickly turn it into a containerized application.
From AI application incubation to production, the Dell APEX Cloud Platform is more than just a playground for virtual assistants: it's an environment where data scientists and developers can collaborate, create, and solve business problems with AI. Dell and our partners continue to deliver hardware and software solutions that can be a launchpad for innovation, and we are committed to provide seamless, powerful, and intuitive tools that transform the way we interact with AI technology.
References
- Digital Assistant with Red Hat OpenShift AI on Dell APEX Cloud Platform for Red Hat OpenShift
- AI Driven Speech Recognition and Synthesis on Dell APEX Cloud Platform for Red Hat OpenShift
- Dell and Red Hat Transform AI Complexity into Opportunity
Author: Bryan McFeeters