Mercurial > ift6266
view scripts/nist_divide.py @ 116:3bec123dd75d
changes on pipeline mecanism: we now sample a different complexity for each transformations, this because when we use the same sampled complexity for all the modules 1/8 of the time we are close to 0 and we obtain an image very close to the source, we now save a complexity for each module in the parameters array
author | Xavier Glorot <glorotxa@iro.umontreal.ca> |
---|---|
date | Wed, 17 Feb 2010 16:22:54 -0500 |
parents | d508f5a8acd0 |
children | 2b6a28e4cadc |
line wrap: on
line source
#!/usr/bin/env python ''' creation des ensembles train, valid et test NIST pur ensemble test est pris tel quel ensemble valid est trainorig[:20000] ensemble train est trainorig[20000:] trainorig est deja shuffled ''' from pylearn.io import filetensor as ft import numpy, os dir1 = "/data/lisa/data/nist/by_class/all/" dir2 = "/data/lisa/data/ift6266h10/" os.system("cp %s %s" % (dir1 + "all_test_data.ft", dir2 + "test_data.ft")) os.system("cp %s %s" % (dir1 + "all_test_labels.ft", dir2 + "test_labels.ft")) f = open(dir1 + "/all_train_data.ft") d = ft.read(f) f = open(dir2 + "valid_data.ft", 'wb') ft.write(f, d[:20000]) f = open(dir2 + "train_data.ft", 'wb') ft.write(f, d[20000:]) f = open(dir1 + "/all_train_labels.ft") d = ft.read(f) f = open(dir2 + "valid_labels.ft", 'wb') ft.write(f, d[:20000]) f = open(dir2 + "train_labels.ft", 'wb') ft.write(f, d[20000:]) for i in ["train", "valid", "test"]: os.chmod(dir2 + i + "_data.ft", 0744) os.chmod(dir2 + i + "_labels.ft", 0744)