GraphNeuralNetworks.jl
GraphNeuralNetworks.jl copied to clipboard
GlobalConcatPool
Concatenates global features to the node features, is used to integrate global properties into the output of the nodes.
Codecov Report
Merging #159 (866e416) into master (eb43bc9) will decrease coverage by
0.00%. The diff coverage is83.33%.
@@ Coverage Diff @@
## master #159 +/- ##
==========================================
- Coverage 85.45% 85.44% -0.01%
==========================================
Files 14 14
Lines 1299 1305 +6
==========================================
+ Hits 1110 1115 +5
- Misses 189 190 +1
| Impacted Files | Coverage Δ | |
|---|---|---|
| src/layers/pool.jl | 93.54% <83.33%> (-2.46%) |
:arrow_down: |
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact),ø = not affected,? = missing dataPowered by Codecov. Last update eb43bc9...866e416. Read the comment docs.
Nice addition, thanks. Needs tests and a few words explaining the behavior in the docstring.
I changed it a bit, so any pooling layer can be used. Do you agree with this change? If so I'll add some documentation
The test uses a GlobalPool layer, is this OK as it happens after the GlobalPool test, or should the test be fully isolated?