This workshop will cover the application of Intel performance optimization tools to AI workloads. The following specific frameworks will be covered:
1. Intel® Optimization for PyTorch. Expected Outcome - understand how to optimize a Pytorch model through intel extension for Pytorch.
2. Intel® Neural Compressor. Expected Outcome - use Intel® Neural Compressor tool to compress AI models based on TensorFlow and PyTorch to speed up inference on Intel® Xeon® CPUs.
3. Intel® Optimizations for TensorFlow. Expected Outcome - be able to see the performance benefit from using Intel Optimizations for Tensorflow over stock Tensorflow framework.
4. Intel® Extension for Scikit-Learn and XGBoost. Expected Outcome – be able to run ML algorithms with Intel Extension for Scikit-learn and compare performance against the original stock version of scikit-learn. Students will see that patching scikit-learn results in a significant increase in performance over the original scikit-learn while also maintaining the same precision.
5. XGBoost: Expected Outcome – analyze the performance benefit from using Intel optimizations upstreamed by Intel to the latest XGBoost compared to un-optimized XGBoost 0.81
Who should attend this class?
The Intel AI Analytics Toolkit gives students, data scientists, AI developers, and researchers familiar Python tools and frameworks to accelerate end-to-end data science and analytics pipelines on Intel® architecture. The components are built using oneAPI libraries for low-level compute optimizations. This toolkit maximizes performance from preprocessing through machine learning and provides interoperability for efficient model development. High level understanding of machine learning and deep learning concepts, and beginner level understanding of various frameworks such as PyTorch, Tensorflow, Scikit-learn and XGBoost will be needed for this workshop.
What will the student, data scientists, AI developers, and researchers get from this class?
By taking this class you will learn Intel optimizations implemented on top of stock versions of data science and AI libraries like NumPy, SciPy, Modin, Scikit Learn, and DL frameworks like Tensorflow and PyTorch, available through the Intel® oneAPI AI Analytics Toolkit. Hands on exercises will be followed to showcase how to get started using Intel AI software and the performance benefits achieved from using Intel optimizations.