Brutex
/
cavis
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Mist gradient check (#57) 

Signed-off-by: AlexDBlack <blacka101@gmail.com>
master
Alex Black 2019-11-19 00:12:59 +11:00 committed by GitHub
parent db7ca956c5
commit 8f96f71f2b
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 1 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

2
deeplearning4j/deeplearning4j-core/src/test/java/org/deeplearning4j/gradientcheck/GradientCheckTests.java
View File

@ -142,7 +142,7 @@ public class GradientCheckTests extends BaseDL4JTest {
// (a) activation function // (a) activation function
// (b) Whether to test at random initialization, or after some learning (i.e., 'characteristic mode of operation') // (b) Whether to test at random initialization, or after some learning (i.e., 'characteristic mode of operation')
// (c) Loss function (with specified output activations) // (c) Loss function (with specified output activations)
Activation[] activFns = {Activation.SIGMOID, Activation.TANH}; Activation[] activFns = {Activation.SIGMOID, Activation.TANH, Activation.MISH};
boolean[] characteristic = {false, true}; //If true: run some backprop steps first boolean[] characteristic = {false, true}; //If true: run some backprop steps first
LossFunction[] lossFunctions = {LossFunction.MCXENT, LossFunction.MSE}; LossFunction[] lossFunctions = {LossFunction.MCXENT, LossFunction.MSE};