I recently had to onboard multiple computer vision datasets each with their own conventions to store camera parameters, depth maps etc. Searching for open source repositories on GitHub that implement data loading for these datasets was extremely helpful. In fact, these days I use code search on GitHub more often than Google search.

Even though I moved some of my personal repos to Codeberg I believe there needs to be a centralised way of searching in open source code.