Enable returning distances from knn and knn_graph #133
Conversation
|
Codecov ReportAll modified and coverable lines are covered by tests ✅
❗ Your organization needs to install the Codecov GitHub app to enable full functionality. Additional details and impacted files@@ Coverage Diff @@
## master #133 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 9 9
Lines 77 116 +39
=========================================
+ Hits 77 116 +39 ☔ View full report in Codecov by Sentry. |
|
This pull request had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. |
|
This pull request had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. |
|
This pull request had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. |
|
This pull request had no activity for 6 months. It will be closed in 2 weeks unless there is some new activity. |
TL;DR Add a switch for
knnandknn_graphto enable returning distances alongside the edge_index.This code has been tested on CPU and CUDA 11.3 with PyTorch 1.11.0 on Ubuntu 18.04.
Related: pyg-team/pytorch_geometric#3010
I still cannot directly decorate
knnandknn_graphwith@torch.jit.scriptper your suggestions (pyg-team/pytorch_geometric#3010 (comment)). So I took a slightly different approach which was similar to pytorch/pytorch#37986. Though@torch.jit.scripthas to be removed from the above two functions, the code that uses those two functions can be correctly jitted. Is it OK with you? @rusty1s