Sotondata
diff --git a/‎code/SdA.py‎
Lines changed: 3 additions & 3 deletions b/‎code/SdA.py‎
Lines changed: 3 additions & 3 deletions
diff --git a/‎doc/SdA.txt‎
Lines changed: 8 additions & 11 deletions b/‎doc/SdA.txt‎
Lines changed: 8 additions & 11 deletions
@@ -280,7 +280,7 @@ def test_score():
 
 
 def test_SdA( finetune_lr = 0.1, pretraining_epochs = 15, \
-              pretrain_lr = 0.05, training_epochs = 1000, \
+              pretrain_lr = 0.001, training_epochs = 1000, \
               dataset='../data/mnist.pkl.gz', batch_size = 1):
     """
     Demonstrates how to train and test a stochastic denoising autoencoder.
@@ -316,7 +316,7 @@ def test_SdA( finetune_lr = 0.1, pretraining_epochs = 15, \
     n_train_batches = train_set_x.value.shape[0] / batch_size
 
     # numpy random generator
-    numpy_rng = numpy.random.RandomState(123)
+    numpy_rng = numpy.random.RandomState(89677)
     print '... building the model'
     # construct the stacked denoising autoencoder class
     sda = SdA( numpy_rng = numpy_rng, n_ins = 28*28, 
@@ -335,7 +335,7 @@ def test_SdA( finetune_lr = 0.1, pretraining_epochs = 15, \
     print '... pre-training the model'
     start_time = time.clock()  
     ## Pre-train layer-wise 
-    corruption_levels = [.1,.1,.0]
+    corruption_levels = [.1,.2,.3]
     for i in xrange(sda.n_layers):
         # go through pretraining epochs 
         for epoch in xrange(pretraining_epochs):
 
@@ -422,19 +422,16 @@ The user can run the code by calling:
 
   python code/SdA.py
 
-By default the code runs 15 pre-training epochs for each layer, with 
-a corruption level of 0.2 and a learning rate of 0.1. Pre-training takes
-76.95 minutes. Fine-tuning is completed after 48 epochs in 115.73 
-minutes and results in a validation score of 1.63 %, with a test 
-performace of 1.68 %. These results were obtained on a machine with an Intel
+By default the code runs 15 pre-training epochs for each layer, with a batch
+size of 1. The  corruption level for the first layer is 0.1, for the second
+0.2 and 0.3 for the third. The pretraining learning rate is was 0.001 and 
+the finetuning learning rate is 0.1. Pre-training takes 585.01 minutes, with 
+an average of 13 minutes per epoch. Fine-tuning is completed after 36 epochs
+in 444.2 minutes, with an average of 12.34 minutes per epoch. The final 
+validation score is 1.39% with a testing score fo 1.3%. 
+These results were obtained on a machine with an Intel
 Xeon E5430 @ 2.66GHz CPU, with a single-threaded GotoBLAS.
 
-For comparison, on a multi-core Intel Xeon X5560 @ 2.80GHz, using multi-threaded MKL
-(with 8 threads), pre-training takes 23.43 minutes and fine-tuning takes 25.61 minutes, 
-meaning a ~4x speed-up at an 8x CPU cost.
-
-Timings accurate as of March 16, 2010.
-
 
 Tips and Tricks
 +++++++++++++++