This folder contains the all the objects in the Cooper-Hewitt collection database exported as individual JSON files. They were created using the export_objects.py script and the add_media.py, both included with this repository.
File names and parent directory structures are generated using the unique ID of the object itself. For example the full path for Piancastelli's "Ceres asks Jupiter for Return of Proserpina" (ID #10001) is: objects/100/01/10001.json
Paths are generated by chopping numeric identifiers in to groups of three (or less) starting front to back. A reference implementation can be found in the utils.py library.
It should be noted that the performance of Git (and GitHub) when dealing with very large repositories is still less than ideal. It appears to be a problem inherent to Git and its indexing processes.
Which is to say: Everything still works but it can be slow. In practice this has the side-effect of making some features of GitHub, like pull requests, nearly impossible because the web application is configured to time out before indexing (of a very large repository) has completed.
There is a good thread about the problem and some potential solutions over here: