For developers focused on deep learning use cases—predictive modeling, recommendation systems, natural language processing, object detection, and tons more—it’s paramount to extract the most workload performance using newer technologies like BF16, graph-level optimizations, and custom kernels.
This session focuses on the performance and ease-of-use benefits for DL training and inference of big models like DLRM (deep learning recommendation model) using Intel® Extension for PyTorch* and Intel® oneAPI Deep Neural Network Library (oneDNN).
Register to hear Senior Deep Learning Engineer Eikan Wang cover:
- Using oneDNN to deliver optimal training and inference workload performance for the PyTorch framework on Intel hardware
- oneDNN-based graph optimizations and custom kernel implementations to boost performance of DLRM modules in PyTorch
- How Intel’s optimized PyTorch extension library can be dynamically loaded as a Python module to offer a more modular design for custom compound operations that are critical to accelerating key DL modules, e.g., the interaction module from DLRM.
Download the software
- Get the Intel® Extension for PyTorch as part of the Intel® oneAPI AI Analytics Toolkit.
- Get oneDNN as part of the Intel® oneAPI Base Toolkit. (Want this tool standalone only? Get it here.)
- Sign up for an Intel® DevCloud for oneAPI account—a free development sandbox with access to the latest Intel® hardware and oneAPI software.
- Explore oneAPI, including developer opportunities and benefits
- Subscribe to the POD—Code Together is an interview series that explores the challenges at the forefront of cross-architecture development. Each bi-weekly episode features industry VIPs who are blazing new trails through today’s data-centric world. Available wherever you get your podcasts.