# HG changeset patch
# User Xavier Glorot <glorotxa@iro.umontreal.ca>
# Date 1269267610 14400
# Node ID 7b4507295ebab3d8e73933e9d347ecee4542c24f
# Parent  44409b6652aa1dcd91d97477f98182a7196bf22e# Parent  f6d9b6b89c2ad01c743a47529152936e9025c5b5
merge

diff -r f6d9b6b89c2a -r 7b4507295eba data_generation/transformations/pycaptcha/Captcha/File.py
--- a/data_generation/transformations/pycaptcha/Captcha/File.py	Mon Mar 22 08:34:48 2010 -0400
+++ b/data_generation/transformations/pycaptcha/Captcha/File.py	Mon Mar 22 10:20:10 2010 -0400
@@ -7,7 +7,7 @@
 # Copyright (C) 2004 Micah Dowty <micah@navi.cx>
 #
 
-import os, random
+import os, random, cPickle
 
 # Determine the data directory. This can be overridden after import-time if needed.
 dataDir = os.path.join(os.path.split(os.path.abspath(__file__))[0], "data")
@@ -41,7 +41,10 @@
             else:
                 path = os.path.join(dataDir, self.basePath, name)
             if os.path.isdir(path):
-                for content in os.listdir(path):
+                f = open(path + '/filelist.pkl')
+                filelist = cPickle.load(f)
+                f.close()
+                for content in filelist:
                     if self._checkExtension(content):
                         paths.append(os.path.join(path, content))
             else:
diff -r f6d9b6b89c2a -r 7b4507295eba data_generation/transformations/ttf2jpg.py
--- a/data_generation/transformations/ttf2jpg.py	Mon Mar 22 08:34:48 2010 -0400
+++ b/data_generation/transformations/ttf2jpg.py	Mon Mar 22 10:20:10 2010 -0400
@@ -10,6 +10,7 @@
 
 import sys, os, fnmatch, random
 import Image, ImageFont, ImageDraw, numpy
+import cPickle
 
 class ttf2jpg():
     def __init__(self, font_file = ''):
@@ -26,8 +27,9 @@
             self.char_list.append(chr(ord('A') + i) )
         for i in range(0,26):
             self.char_list.append(chr(ord('a') + i) )
-        files = os.listdir(self.font_dir)
-        self.font_files = fnmatch.filter(files, '*.ttf') + fnmatch.filter(files, '*.TTF')
+        f = open( self.font_dir + 'filelist.pkl' ,'r')
+        self.font_files = cPickle.load(f)
+        f.close()
 
     # get font name
     def get_settings_names(self):
diff -r f6d9b6b89c2a -r 7b4507295eba deep/stacked_dae/__init__.py
diff -r f6d9b6b89c2a -r 7b4507295eba deep/stacked_dae/nist_sda.py
diff -r f6d9b6b89c2a -r 7b4507295eba deep/stacked_dae/sgd_optimization.py
diff -r f6d9b6b89c2a -r 7b4507295eba deep/stacked_dae/stacked_dae.py
diff -r f6d9b6b89c2a -r 7b4507295eba deep/stacked_dae/utils.py
diff -r f6d9b6b89c2a -r 7b4507295eba scripts/launch_generate100.py
--- a/scripts/launch_generate100.py	Mon Mar 22 08:34:48 2010 -0400
+++ b/scripts/launch_generate100.py	Mon Mar 22 10:20:10 2010 -0400
@@ -3,12 +3,13 @@
 import os
 dir1 = "/data/lisa/data/ift6266h10/"
 
-mach = "maggie16.iro.umontreal.ca,maggie15.iro.umontreal.ca"
+mach = ["maggie16.iro.umontreal.ca,zappa8@iro.umontreal.ca"]
 
+#test and valid sets
 for i,s in enumerate(['valid','test']):
     for j,c in enumerate([0.3,0.5,0.7,1]):
         l = str(c).replace('.','')
-        os.system("dbidispatch --condor --os=fc4,fc7,fc9 --machine=%s ./run_pipeline.sh -o %sdata/P%s_%s_data.ft -p %sdata/P%s_%s_params -x %sdata/P%s_%s_labels.ft -f %s%s_data.ft -l %s%s_labels.ft -c %socr_%s_data.ft -d %socr_%s_labels.ft -m 0.3 -z 0.1 -a 0.1 -b 0.25 -g 0.25 -s %d -y %d" % (mach, dir1, l, s, dir1, l, s, dir1, l, s, dir1, s, dir1, s, dir1, s, dir1, s, [20000,80000][i], 200+i*4+j))
+        os.system("dbidispatch --condor --os=fc4,fc7,fc9 --machine=%s ./run_pipeline.sh -o %sdata/P%s_%s_data.ft -p %sdata/P%s_%s_params -x %sdata/P%s_%s_labels.ft -f %s%s_data.ft -l %s%s_labels.ft -c %socr_%s_data.ft -d %socr_%s_labels.ft -m %s -z 0.1 -a 0.1 -b 0.25 -g 0.25 -s %d -y %d" % (mach, dir1, l, s, dir1, l, s, dir1, l, s, dir1, s, dir1, s, dir1, s, dir1, s, c ,[20000,80000][i], 200+i*4+j))
 
 #P07
 for i in range(100):
@@ -17,3 +18,9 @@
 #PNIST07
 for i in range(100):
     os.system("dbidispatch --condor --os=fc4,fc7,fc9 --machine=%s ./run_pipeline.sh -o %sdata/PNIST07_train%d_data.ft -p %sdata/PNIST07_train%d_params -x %sdata/PNIST07_train%d_labels.ft -f %strain_data.ft -l %strain_labels.ft -c %socr_train_data.ft -d %socr_train_labels.ft -m 0.7 -z 0.1 -a 0.1 -b 0.25 -g 0.25 -s 819200 -y %d -t %d" % (mach, dir1, i, dir1, i, dir1, i, dir1, dir1, dir1, dir1, 100+i,1))
+
+
+
+#P07
+#for i in [90,94]:#[2,10,13,15,20,49,68,82,86,90,94]:
+   #os.system("dbidispatch --condor --mem=3900 --os=fc4,fc7,fc9 --machine=maggie16.iro.umontreal.ca --machine=maggie15.iro.umontreal.ca --machine=zappa8@iro.umontreal.ca ./run_pipeline.sh -o %sdata2/P07_train%d_data.ft -p %sdata2/P07_train%d_params -x %sdata2/P07_train%d_labels.ft -f %strain_data.ft -l %strain_labels.ft -c %socr_train_data.ft -d %socr_train_labels.ft -m 0.7 -z 0.1 -a 0.1 -b 0.25 -g 0.25 -s 819200 -y %d" % (dir1, i, dir1, i, dir1, i, dir1, dir1, dir1, dir1,100+i))