There's no anchor so search for
divideby in this page.
The documentation says:
The value by which to scale the input data before it is inserted into the network.
When the data is pictures with the pixels ranging from 0 to 255, one of the best practices is to normalize it to a [0, 1) range. Another would be z-score normalization (remove the mean, then divide by stdev). But that's a different discussion altogether :)