Mercurial > pylearn

changeset 404:8cc11ac97087

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Debugging simple AA a bit
author Joseph Turian <turian@gmail.com>
date Thu, 10 Jul 2008 00:51:32 -0400
parents 273e5c03003e
children be4209cd568f
files sandbox/rbm/globals.py sandbox/simple_autoassociator/graph.py sandbox/simple_autoassociator/model.py sandbox/simple_autoassociator/parameters.py
diffstat 4 files changed, 9 insertions(+), 7 deletions(-) [+]
line wrap: on
line diff
--- a/sandbox/rbm/globals.py	Wed Jul 09 17:55:46 2008 -0400
+++ b/sandbox/rbm/globals.py	Thu Jul 10 00:51:32 2008 -0400
@@ -4,8 +4,8 @@
 
 INPUT_DIMENSION = 10
 #INPUT_DIMENSION = 100
-#HIDDEN_DIMENSION = 100
-HIDDEN_DIMENSION = 10
+HIDDEN_DIMENSION = 100
+#HIDDEN_DIMENSION = 10
 #HIDDEN_DIMENSION = 6
 LEARNING_RATE = 0.1
 LR = LEARNING_RATE
--- a/sandbox/simple_autoassociator/graph.py	Wed Jul 09 17:55:46 2008 -0400
+++ b/sandbox/simple_autoassociator/graph.py	Thu Jul 10 00:51:32 2008 -0400
@@ -17,10 +17,10 @@
 loss_unsummed = binary_crossentropy(y, x)
 loss = t.sum(loss_unsummed)
 
-(gw1, gb1, gw2, gb2, gy) = t.grad(loss, [w1, b1, w2, b2, y])
+(gw1, gb1, gw2, gb2, gy, gh) = t.grad(loss, [w1, b1, w2, b2, y, h])
 
 import theano.compile
 
 inputs  = [x, w1, b1, w2, b2]
-outputs = [y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy]
+outputs = [y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy, gh]
 trainfn = theano.compile.function(inputs, outputs)
--- a/sandbox/simple_autoassociator/model.py	Wed Jul 09 17:55:46 2008 -0400
+++ b/sandbox/simple_autoassociator/model.py	Thu Jul 10 00:51:32 2008 -0400
@@ -28,13 +28,14 @@
         for idx in instance.keys():
             x[idx] = instance[idx]
 
-        (y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy) = trainfn(x, self.parameters.w1, self.parameters.b1, self.parameters.w2, self.parameters.b2)
+        (y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy, gh) = trainfn(x, self.parameters.w1, self.parameters.b1, self.parameters.w2, self.parameters.b2)
         print
         print "instance:", instance
         print "x:", x
         print "OLD y:", y
         print "OLD loss (unsummed):", loss_unsummed
         print "gy:", gy
+        print "gh:", gh
         print "OLD total loss:", loss
         print "gw1:", gw1
         print "gb1:", gb1
@@ -48,7 +49,7 @@
         self.parameters.b2  -= LR * gb2
 
         # Recompute the loss, to make sure it's descreasing
-        (y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy) = trainfn(x, self.parameters.w1, self.parameters.b1, self.parameters.w2, self.parameters.b2)
+        (y, h, loss, loss_unsummed, gw1, gb1, gw2, gb2, gy, gh) = trainfn(x, self.parameters.w1, self.parameters.b1, self.parameters.w2, self.parameters.b2)
         print "NEW y:", y
         print "NEW loss (unsummed):", loss_unsummed
         print "gy:", gy
--- a/sandbox/simple_autoassociator/parameters.py	Wed Jul 09 17:55:46 2008 -0400
+++ b/sandbox/simple_autoassociator/parameters.py	Thu Jul 10 00:51:32 2008 -0400
@@ -20,7 +20,8 @@
             self.w1 = (numpy.random.rand(input_dimension, hidden_dimension)-0.5)/input_dimension
             self.w2 = (numpy.random.rand(hidden_dimension, input_dimension)-0.5)/hidden_dimension
             self.b1 = numpy.zeros(hidden_dimension)
-            self.b2 = numpy.zeros(input_dimension)
+            #self.b2 = numpy.zeros(input_dimension)
+            self.b2 = numpy.array([10, 0, 0, -10])
         else:
             self.w1 = numpy.zeros((input_dimension, hidden_dimension))
             self.w2 = numpy.zeros((hidden_dimension, input_dimension))