Introduction to IBM Watson Machine Learning Accelerator
WML Accelerator provides an end-to-end, deep learning platform for data scientists. This includes the complete lifecycle management from installation and configuration; data ingest and preparation; building, optimizing, and distributing the training model; to moving the model into production. WML Accelerator enables you to iterate through the training cycle on more data to continuously improve the model over time.
WML Accelerator provides many optimizations that accelerate performance, improve resource utilization, and reduce installation, configuration, and management complexities, such as the following:
- Distributed deep learning architecture that simplifies the process of training deep learning models across a cluster for faster time to results.
- Large Model Support seamlessly increases the amount of memory available for deep learning models beyond the 16 GB or 32 GB footprint of your GPU memory, enabling more complex models with larger, more high-resolution data inputs.
- Enhanced data ingest, preparation, and transformation tools, using Apache Spark for data management.
- IBM® Power® Systems servers architected for artificial intelligence (AI) applications, incorporating high-bandwidth and low-latency NVLink connections between GPU accelerators for peer-to-peer communications, and directly connecting GPU accelerators to system CPUs (and system memory).
- Powerful model development tools, including real-time training visualization and runtime monitoring of accuracy and hyper-parameter search and optimization, for faster model development.
- Ready-to-use deep learning frameworks (TensorFlow, PyTorch, and IBM Caffe) are included.
- Multitenant architecture designed to run deep learning, high-performance analytics, and other long-running services and frameworks on shared resources.