Skip to content

GNNlib-v1.4.1

Latest

Choose a tag to compare

@github-actions github-actions released this 04 Nov 18:24
· 5 commits to master since this release

GNNlib GNNlib-v1.4.1

Diff since GNNlib-v1.4.0

see the CHANGELOG

Merged pull requests:

  • Update AMDGPU requirement from 1 to 1, 2.7 (#698) (@dependabot[bot])
  • docs: fix some typos (#699) (@jeis4wpi)
  • Fix Pkg.test failure for registered packages (#700) (#701) (@CarloLucibello)
  • Fix Enzyme differentiation of GCNConv/SGConv/TAGConv on adjacency-matrix graphs via a keyword-free COO conversion helper (#703) (@Parvm1102)
  • Add GNNGraphsMooncakeExt.jl (#704) (@Parvm1102)
  • Bump JuliaRegistries/TagBot from 1.25.10 to 1.25.11 (#705) (@dependabot[bot])
  • Add GNNGraphsEnzymeCoreExt marking scaled_laplacian inactive for Enzyme (#706) (@Parvm1102)
  • Slice GNNRecurrence input by indexing instead of eachslice views (#707) (@Parvm1102)
  • Bump JuliaRegistries/TagBot from 1.25.11 to 1.25.12 (#711) (@dependabot[bot])
  • Bump JuliaRegistries/TagBot from 1.25.12 to 1.25.14 (#714) (@dependabot[bot])
  • Let Lux GNNChain return states of the same structure as the input (#715) (@simsurace)
  • Pass the edge weights in the GPU w_mul_xj fallback (#718) (@simsurace)
  • Require MLDatasets 0.7 in the test and docs environments (#719) (@simsurace)
  • Fall back to synthetic data when a tutorial dataset fails to download (#720) (@simsurace)
  • Create the eigensolver start vector on the device of the matrix (#722) (@simsurace)
  • Bump JuliaRegistries/TagBot from 1.25.14 to 1.25.15 (#723) (@dependabot[bot])
  • Bring the GNNLux tutorials to parity with GNN (#724) (@simsurace)

Closed issues:

  • Proposal: Add comprehensive Link Prediction tutorial to documentation. (#669)
  • copy_xj mean and e_mul_xj mean rules for Mooncake (#689)
  • TagBot: Manual intervention needed for releases (#694)
  • Testing GraphNeuralNetworks fails (#700)