Generating images from a text description is one of the most impressive applications right now. Below, you will find some exemplary papers released mainly by large companies. These can be seen as case studies on how to set up “realistic” large-scale models for complex generation tasks. Interesting questions to ask include:
Even though most people do not have the resources available to recreate these kinds of models, there is still a lot to learn on scalable, efficient deep learning.
You are not expected to read all these papers in detail, but consider “academic speed reading”: Read the abstract, introduction, and conclusion. Check the figures and tables for main results. Done! Alternatively, you could try reading one or two papers in detail. Most of these will also have higher-level blog posts accompanying the release, which you should find via internet search.