TastyPancakes/Flux.jl
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

pull out graph-based affine layer

Browse Source
This commit is contained in:
Mike J Innes 2017-08-19 20:26:27 +01:00
parent b18d8cd08f
commit c7a07562d0
1 changed files with 11 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
test/compiler.jl
View File

@ -1,8 +1,17 @@
using DataFlow, MacroTools using DataFlow, MacroTools
using Flux: Affine, Param, Recurrent, squeeze, unsqueeze, stack using Flux: Param, Recurrent, squeeze, unsqueeze, stack
using Flux.Compiler: @net, graph using Flux.Compiler: @net, graph
using DataFlow: Line, Frame using DataFlow: Line, Frame
@net type Affine
W
b
x -> x*W .+ b
end
Affine(in::Integer, out::Integer; init = Flux.initn) =
Affine(init(in, out), init(1, out))
@net type TLP @net type TLP
first first
second second
@ -42,7 +51,7 @@ let tlp = TLP(Affine(10, 21), Affine(20, 15))
e e
end end
@test e.trace[end].func == :TLP @test e.trace[end].func == :TLP
@test e.trace[end-1].func == Symbol("Flux.Affine") @test e.trace[end-1].func == Symbol("Affine")
end end
function apply(model, xs, state) function apply(model, xs, state)