New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Feature] Fixed sampler with limit on sampled nodes/edges in batch subgraph #6668
base: master
Are you sure you want to change the base?
Conversation
Fix typo in ShaDowKHopSampler sample() function
Merge latest DGL changes
Merge latest DGL changes
Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:
|
Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:
|
Thanks for the edits, @frozenbugs. Happy to address all the formatting errors. Do you see any issues with the function implementation itself? |
Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:
|
Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:
|
the code LGTM now, can you run The test case can be put here: tests/python/pytorch/dataloading, and you can use this test case as example: https://github.com/dmlc/dgl/blob/master/tests/python/pytorch/graphbolt/impl/test_in_subgraph_sampler.py |
Not authorized to trigger CI. Please ask core developer to help trigger via issuing comment:
|
|
@ayushnoori if you don't plan to write a unit test, we can also merge this, can you add a comment in capped_neighbor_sampler.py, says "This code was contributed by a community member (ayushnoori). There aren't currently any unit tests in place to verify its functionality, so please be cautious if you need to make any changes to the code's logic." |
Description
Subgraph sampler that sets an upper bound on the number of nodes included in each layer of the sampled subgraph. At each layer, the frontier is randomly subsampled. Rare node types can also be upsampled by taking the scaled square root of the sampling probabilities (best strategy TBD). The new
FixedSampler
performs node-wise neighbor sampling and returns the subgraph induced by all the sampled nodes.The relevant issue is: #6623, thanks to @frozenbugs and @jermainewang for their input and review.
Checklist
Please feel free to remove inapplicable items for your PR.
Please note that unit tests for
FixedSampler
are not yet written.