IBM PowerAI Base is a software suite that includes open source machine learning frameworks such as Scikit-Learn and DL frameworks such as TensorFlow, PyTorch, and Caffe. It includes a variety of features to increase performance and reduce training time for DL models.
In a more comprehensive business version, the IBM Watson Machine Learning Accelerator, the AI solution combines common open source frameworks for DL with efficient AI development tools and accelerated, unrestricted IBM Power Systems servers and makes them readily available. IBM support is also included.
The platform provides data scientists with the ability to create and manage Spark clusters, sharing the underlying resources of the IBM Watson Machine Learning Accelerator servers, including memory, CPUs, and GPUs.
The PowerAI Distributed Deep Learning and extensive model support allows for faster results. Model development tools include real-time visualization of training, accuracy monitoring at runtime, and hyper-parameter search and optimization. The Distributed DL can scale to thousands of nodes.
The Watson Machine Learning Accelerator software package includes Anaconda Libraries, IBM PowerAI Base, and IBM Spectrum Conductor Deep Learning Impact (DLI) - see also Figure 1.
The platform can be enriched, among others, with IBM PowerAI Vision.