Maximum number of neighbours in the imdb tutorial

[JanSellner]

There is an inconsistency regarding the maximum number of neighbours to use in the tutorial. In the Augment training data with graph neighbors section, a maximum of 1 neighbour is mentioned. However, in the following run of the pack_nbrs tool the parameter --max_nbrs=3 is used. The HParams object later specifies self.num_neighbors = 2.

If I had to guess, I would say it should be 2 neighbours everywhere. Please note that this also changes the output of the pack_nbrs tool.

Thanks for sharing this library!

[arjung]

Thanks @Milania1 for running through our tutorial and for reporting the inconsistency! The only typo is in the text of the Augment training data with graph neighbors section (it should be 3 instead of 1). I will fix this shortly.

The value of HParams.num_neighbors can be any value in the inclusive interval [0,3]. This is because the input training data for graph-based NSL was generated (by the pack_nbrs program) to include a maximum of 3 neighbors per sample. So, when training a model, we may choose to use fewer neighbors per sample. The IMDB tutorial illustrates using 2 neighbors per sample but you may also use 0, 1, or 3 neighbors per sample. Hope this clarifies your question.

[arjung]

Fixed in 6cb6b22.