Update GNNChain #202
Merged
Update GNNChain #202
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
CarloLucibello
commented
Jul 22, 2022
Comment on lines
+108
to
+115
| ## TODO see if this is faster for small chains | ||
| ## see https://github.com/FluxML/Flux.jl/pull/1809#discussion_r781691180 | ||
| # @generated function _applychain(layers::Tuple{Vararg{<:Any,N}}, g::GNNGraph, x) where {N} | ||
| # symbols = vcat(:x, [gensym() for _ in 1:N]) | ||
| # calls = [:($(symbols[i+1]) = _applylayer(layers[$i], $(symbols[i]))) for i in 1:N] | ||
| # Expr(:block, calls...) | ||
| # end | ||
| # _applychain(layers::NamedTuple, g, x) = _applychain(Tuple(layers), x) |
There was a problem hiding this comment.
note to myself: remember to benchmark this before merging
There was a problem hiding this comment.
Benchmarked on a small graph / net
n, deg = 10, 4
din, d, dout = 10, 3, 4, 2
g = GNNGraph(random_regular_graph(n, deg),
graph_type=GRAPH_T,
ndata= randn(Float32, din, n))
x = g.ndata.x
gnn = GNNChain(GCNConv(din => d),
BatchNorm(d),
x -> tanh.(x),
GraphConv(d => d, tanh),
Dropout(0.5),
Dense(d, dout))There is a performance increase with the generated _applychain but not large enough for the change to be worthwhile
julia> using BenchmarkTools
### without @generated _applychain
julia> @btime gnn(g, x)
7.469 μs (84 allocations: 12.48 KiB)
2×10 Matrix{Float32}:
-0.8186 -0.570312 -0.777638 … -0.641642 -0.684857 -0.975505
0.305567 0.559996 0.631279 0.4687 0.479899 0.321139
julia> @btime gradient(x -> sum(gnn(g, x)), x)
515.917 μs (2422 allocations: 160.52 KiB)
(Float32[0.3974119 -0.5917164 … -0.9200875 1.1957061; -0.54502636 -1.5056851 … -2.6915464 2.5114572; -0.97105116 0.7726713 … 1.0995824 -1.5013595],)
### with @generated _applychain
julia> @btime gnn(g, x)
6.825 μs (73 allocations: 11.55 KiB)
2×10 Matrix{Float32}:
-0.8186 -0.570312 -0.777638 … -0.641642 -0.684857 -0.975505
0.305567 0.559996 0.631279 0.4687 0.479899 0.321139
julia> @btime gradient(x -> sum(gnn(g, x)), x)
454.750 μs (2157 allocations: 161.00 KiB)
(Float32[-0.564121 0.3105453 … 0.19531891 -0.22819248; -0.6428803 0.13550264 … 0.9421329 -0.79201597; 0.7816532 -0.4734739 … 0.23667078 0.033573348],)In both cases the gradient is very slow, this should be further investigated
Codecov Report
@@ Coverage Diff @@
## master #202 +/- ##
==========================================
+ Coverage 86.44% 87.35% +0.90%
==========================================
Files 15 15
Lines 1365 1368 +3
==========================================
+ Hits 1180 1195 +15
+ Misses 185 173 -12
Continue to review full report at Codecov.
|
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Keeping in sync GNNChain with last months' changes with Flux.Chain implementation (FluxML/Flux.jl#1809)
TODO: