Skip to content

Test GCN edge weight gradients through GNNChain - #738

Open
DHAKED440 wants to merge 1 commit into
JuliaGraphs:masterfrom
DHAKED440:pr/issue-443-edge-weight-gradient-test
Open

DHAKED440 wants to merge 1 commit into
JuliaGraphs:masterfrom
DHAKED440:pr/issue-443-edge-weight-gradient-test

Conversation

@DHAKED440

@DHAKED440 DHAKED440 commented Nov 16, 2026 •

Copy link
Copy Markdown
Contributor

Summary
Adds regression coverage for trainable edge weights in GCNConv when used through GNNChain.
Changes
Adds a test in GraphNeuralNetworks/test/layers/conv.jl that checks:

  • Gradients are finite and nonzero for edge weights and both GCN weight matrices.
  • Changing edge weights changes the model output.
  • Updating edge weights or network parameters changes the output.
    This is a 43-line test-only change. It does not modify package source code or public APIs.
    Tests
    The existing GCNConv test and the new regression test passed: 89/89 tests. git diff --check also passed. The full package test suite was not run.

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant