Cost-effective Inference for Ranker Models

Modern recommender system models can be very compute- and memory-intensive and expensive to serve. In this post, we’ll go over some of the things we did at…

This is an excerpt. The full article continues at ShareChat.