Workshop: How to Optimize Deep Learning Models for Production

Topics covered: 1. What model pruning is, including benefits and downsides; 2. SOTA pruning algorithms and techniques that you can implement today; 3. SparseML, an open-source tool that makes pruning easy and successful; 4. Guaranteed ways to get production performance out of a pruned model. After watching this video, you’ll be able to optimize your NLP and/or computer vision model, apply your own data with a few lines of code, and deploy it on commodity CPUs at GPU-level speeds. *Accompanying slides are linked below. Timeline: 00:00 - Intro & agenda 02:45 - Advantages and disadvantages of model optimizations 05:00 - SOTA optimization algorithms and techniques 09:25 - Ways to apply optimizations to a model 11:25 - Available model optimization tools 13:27 - Intro to SparseML and SparseZoo, tools we’ll be using during the workshop 15:25 - NLP example using BERT - how to apply optimizations to NLP models with ease using recipes 35:25 - Computer vision example using YOLOv5 - how to apply optimizations to com