Now parameter num_neighbors is fixed in cool_graph/config/training/default.yaml:
or in cool_graph/config/training/in_memory_data.yaml:
But these parameters are not optimal for all datasets.
It is better to have flexible algorithm for sampling neighbors.
For example, we could limit the number of neighbors by 95%-quantile of node degree, or less for 2-level neighbors
Now parameter
num_neighborsis fixed in cool_graph/config/training/default.yaml:or in cool_graph/config/training/in_memory_data.yaml:
But these parameters are not optimal for all datasets.
It is better to have flexible algorithm for sampling neighbors.
For example, we could limit the number of neighbors by 95%-quantile of node degree, or less for 2-level neighbors