[Diff since v0.6.13](CarloLucibello/GraphNeuralNetworks.jl@v0.6.13...…

…v0.6.14)

**Merged pull requests:**
- batch wider eltype (CarloLucibello#340) (@CarloLucibello)
- Remove type constraints on Flux.batch for GNNHeteroGraph  (CarloLucibello#342) (@AarSeBail)

**Closed issues:**
- The constraint in Flux.batch(gs::AbstractVector{<:GNNHeteroGraph}) does not seem to be strong enough (CarloLucibello#341)

[Diff since v0.6.12](CarloLucibello/GraphNeuralNetworks.jl@v0.6.12...…

…v0.6.13)

**Closed issues:**
- Edge weights not properly documented for `GNNHeteroGraph`s (and implement new function to add new edge weights?) (CarloLucibello#331)

**Merged pull requests:**
- Bump actions/checkout from 3 to 4 (CarloLucibello#336) (@dependabot[bot])
- improve add_edges for graph and heterographs (CarloLucibello#337) (@CarloLucibello)
- extend reduce_nodes and graph_indicator for heterographs (CarloLucibello#339) (@CarloLucibello)

[Diff since v0.6.11](CarloLucibello/GraphNeuralNetworks.jl@v0.6.11...…

…v0.6.12)

**Closed issues:**
- Update `Flux.trainable` (CarloLucibello#323)
- HeteroGraphConv bug: ERROR: duplicate field name in NamedTuple: "movie" is not unique (CarloLucibello#332)
- `add_edges` for GNNHeteroGraph does not allow providing the number of nodes (CarloLucibello#334)

**Merged pull requests:**
- Update `Flux.trainable`  (CarloLucibello#330) (@aurorarossi)
- Fix heteroconv aggregation + tests (CarloLucibello#333) (@svilupp)
- fixes add_edges + adds tests (CarloLucibello#335) (@svilupp)

[Diff since v0.6.10](CarloLucibello/GraphNeuralNetworks.jl@v0.6.10...…

…v0.6.11)

**Closed issues:**
- `AGNNConv` behaviour different from mathematical definition (due to self loops) (CarloLucibello#325)

**Merged pull requests:**
- Add temporal graph documentation (CarloLucibello#327) (@aurorarossi)
- fix AGNNConv docstring and add other options (CarloLucibello#328) (@CarloLucibello)

[Diff since v0.6.9](CarloLucibello/GraphNeuralNetworks.jl@v0.6.9...v0…

….6.10)

**Merged pull requests:**
- Add `TGCNCell` layer (CarloLucibello#314) (@aurorarossi)
- Add `Flux.Recur` and `TGCN` (CarloLucibello#319) (@aurorarossi)
- Fix typos and `julia-repl` text (CarloLucibello#320) (@aurorarossi)
- Add traffic prediction example (CarloLucibello#321) (@aurorarossi)
- Add `A3TGCN`  model (CarloLucibello#322) (@aurorarossi)
- Add traffic prediction tutorial (CarloLucibello#324) (@aurorarossi)
- Julia repl rendering fix (CarloLucibello#326) (@aurorarossi)

[Diff since v0.6.8](CarloLucibello/GraphNeuralNetworks.jl@v0.6.8...v0…

….6.9)

**Closed issues:**
- use extension instead of CUDA hard dependence (CarloLucibello#317)

**Merged pull requests:**
- Remove CUDA dependence in favor of extension (CarloLucibello#318) (@CarloLucibello)

[Diff since v0.6.7](CarloLucibello/GraphNeuralNetworks.jl@v0.6.7...v0…

….6.8)

**Closed issues:**
- Implement Learnable Structural Positional Encoding (LSPE) (CarloLucibello#190)
- add docs for HeteroGraphConv (CarloLucibello#302)
- add batching for GNNHeteroGraph (CarloLucibello#303)

**Merged pull requests:**
- Add random temporal hyperbolic graph (CarloLucibello#313) (@aurorarossi)
- move to Flux 0.14 (CarloLucibello#316) (@CarloLucibello)

[Diff since v0.6.6](CarloLucibello/GraphNeuralNetworks.jl@v0.6.6...v0…

….6.7)

**Closed issues:**
- Convolutions for GNNHeteroGraphs (CarloLucibello#254)
- need more informative error for dimension mismatch (CarloLucibello#283)
- Local pooling in graph regression/classification problems (CarloLucibello#307)

**Merged pull requests:**
- Improve error description of `setproperty!` (CarloLucibello#287) (@aurorarossi)
- Fix spacing in documentation example (CarloLucibello#289) (@asinghvi17)
- Fix typos (CarloLucibello#290) (@aurorarossi)
- Add TemporalSnapshotsGNNGraph struct (CarloLucibello#293) (@aurorarossi)
- Add docstrings for TemporalSnapshotsGNNgraph (CarloLucibello#294) (@aurorarossi)
- modernize examples and create example notebooks (CarloLucibello#295) (@CarloLucibello)
- make `add_snapshot` non mutating (CarloLucibello#296) (@CarloLucibello)
- Add getproperty function to TemporalSnapshotsGNNGraph (CarloLucibello#297) (@aurorarossi)
- [heterograph] add has_edge, get_index, add_edges (CarloLucibello#298) (@CarloLucibello)
- HeteroGraphConv implementation (CarloLucibello#300) (@CarloLucibello)
- update examples to explicit gradient (CarloLucibello#301) (@CarloLucibello)
- improvement for differentiability with heterogeneous graphs (CarloLucibello#304) (@CarloLucibello)
- Add Temporal random radius graph generator (CarloLucibello#305) (@aurorarossi)
- CompatHelper: bump compat for NNlib to 0.9, (keep existing compat) (CarloLucibello#306) (@github-actions[bot])
- Flux.batch for GNNHeteroGraph (CarloLucibello#309) (@CarloLucibello)
- heterograph guide (CarloLucibello#310) (@CarloLucibello)
- HeteroGraphConv documentation (CarloLucibello#312) (@CarloLucibello)

[Diff since v0.6.5](CarloLucibello/GraphNeuralNetworks.jl@v0.6.5...v0…

….6.6)

**Closed issues:**
- `add_edges` adds a non-existent edge to its `DataStore` (CarloLucibello#280)

**Merged pull requests:**
- Fix few typos in tutorials (CarloLucibello#278) (@natema)
- CompatHelper: bump compat for StatsBase to 0.34, (keep existing compat) (CarloLucibello#279) (@github-actions[bot])
- Fix a few typos (CarloLucibello#281) (@pitmonticone)
- fix cat empty features (CarloLucibello#286) (@CarloLucibello)

[Diff since v0.6.4](CarloLucibello/GraphNeuralNetworks.jl@v0.6.4...v0…

….6.5)

**Merged pull requests:**
- Add random_walk_pe (CarloLucibello#273) (@aurorarossi)
- add Set2Set (CarloLucibello#276) (@CarloLucibello)