Reproducing reported FLOP numbers

[HaniItani]

Hello @zeliu98,

Thank you for sharing your work. Can you please provide details or code snippet to reproduce the FLOP metric reported in the main paper in Table 1? Is it reported over a fixed number of points (N) and number of neighbors (k), or is it averaged over the entire inference? I am using the Tensorflow version.

Much appreciated!