For ML-powered advertisement systems, when building a big model to serve all the different businesses and channels, it is really important to make your model aware of the distinction between different domains. Here we present two papers from Alibaba which solve the problem of multi-domain learning in the field of online ads.

This is a tutorial for beginners trying to deploy their PyTorch models online and build a model inference service.

This paper from Google Research shows us how a Transformer can be utilized to “remember” the indexing of documents via a seq2seq fashion of training and decoding.

We built a new search engine for papers published on ICCV and CVPR in recent years, which has been deployed on AWS and can be accessed via this link. Full code is available on our Github.

This post covers 6 types of algorithms on graph that people may encounter when they are doing leetcode for graph-related problems.