Skip to content
This repository was archived by the owner on May 21, 2022. It is now read-only.

Commit 4ef34a0

Browse files
authored
Merge pull request #5 from rdeits/fix-tests
turn off InputNorm (which is broken anyway) so package can be imported
2 parents aa05015 + 8fba4c5 commit 4ef34a0

File tree

2 files changed

+5
-2
lines changed

2 files changed

+5
-2
lines changed

src/Transformations.jl

+2-2
Original file line numberDiff line numberDiff line change
@@ -14,7 +14,7 @@ import LearnBase: transform, transform!,
1414
learn!
1515
# import StatsBase: logistic, logit
1616
import OnlineStats
17-
import OnlineStats: weight, Weight, BoundedEqualWeight, Variances
17+
import OnlineStats: weight, Weight, BoundedEqualWeight
1818

1919
export
2020
input_node,
@@ -173,7 +173,7 @@ include("linear.jl")
173173
include("affine.jl")
174174
include("misc.jl")
175175
include("layernorm.jl")
176-
include("inputnorm.jl")
176+
# include("inputnorm.jl")
177177
include("activations.jl")
178178
include("whiten.jl")
179179
include("chain.jl")

src/inputnorm.jl

+3
Original file line numberDiff line numberDiff line change
@@ -2,6 +2,9 @@
22
#=
33
Similar to Batch Normalization, except online and without the rescaling/skew
44
y = (a .- μ) ./ σ
5+
6+
TODO: This is currently broken because OnlineStats.Variances no longer
7+
exists.
58
=#
69
type InputNorm{T,W<:Weight} <: Transformation
710
n::Int # maps n --> n

0 commit comments

Comments
 (0)