Learn How to Apply Second-Order Pruning Algorithms for SOTA Model Compression
At Neural Magic, we have developed state-of-the-art research for enabling faster and more efficient models through second-order pruning methods. These include Woodfisher (NeurIPS 2020), MFAC (NeurIPS 2021), and oBERT (EMNLP 2022).
Second-order pruning methods enable higher sparsity while maintaining accuracy by removing weights that directly affect the loss function the least. The end result is a sparse model with much smaller files, lower latency, and higher throughput. For example, by applying the methods above, a ResNet-50 image classification model can be pruned 95% while maintaining 99% of the baseline accuracy, decreasing the files size from the original 90.3 MB to 9.3 MB
In this 30-minute webinar (plus 15-minute Q&A!), we will walk through the research, production results, and intuition for how these algorithms work. And most importantly, our experts will run through how to apply second-order pruning algorithms for SOTA model compression to your current ML projects!
- Mark Kurtz, Director of Machine Learning, Neural Magic
- Eldar Kurtić, Research Consultant, Neural Magic
Event Date: April 6, 2023
Event Time: 1:00 PM EST; 10:00 AM PST
Let us know you are coming. Fill out the form below!