def run():
disturb_mem.disturb_mem()
b = sharedX(np.zeros((2,)))
channels = OrderedDict()
disturb_mem.disturb_mem()
v_max = b.max(axis=0)
v_min = b.min(axis=0)
v_range = v_max - v_min
updates = []
for i, val in enumerate([
v_max.max(),
v_max.min(),
v_range.max(),
]):
disturb_mem.disturb_mem()
s = sharedX(0., name='s_'+str(i))
updates.append((s, val))
for var in theano.gof.graph.ancestors(update for var, update in updates):
if var.name is not None:
if var.name[0] != 's' or len(var.name) != 2:
var.name = None
for key in channels:
updates.append((s, channels[key]))
file_path='nondeterminism_6.txt'
mode = RecordMode(file_path=file_path,
replay=0)
f = theano.function([], mode=mode, updates=updates, on_unused_input='ignore', name='f')
"""
print 'type(f): ',type(f)
print 'elements of f:'
for elem in dir(f):
print '\t',elem
print 'type(f.fn): ',type(f.fn)
print 'elements of f.fn:'
for elem in dir(f.fn):
print '\t',elem
"""
trials = 1
for i in xrange(trials):
disturb_mem.disturb_mem()
f()
mode.record.f.flush()
mode.record.f.close()
mode.set_record(Record(file_path=file_path, replay=1))
for i in xrange(trials):
disturb_mem.disturb_mem()
f()
def run():
disturb_mem.disturb_mem()
b = sharedX(np.zeros((2, )))
channels = OrderedDict()
disturb_mem.disturb_mem()
v_max = b.max(axis=0)
v_min = b.min(axis=0)
v_range = v_max - v_min
updates = []
for i, val in enumerate([
v_max.max(),
v_max.min(),
v_range.max(),
]):
disturb_mem.disturb_mem()
s = sharedX(0., name='s_' + str(i))
updates.append((s, val))
for var in theano.gof.graph.ancestors(update for var, update in updates):
if var.name is not None:
if var.name[0] != 's' or len(var.name) != 2:
var.name = None
for key in channels:
updates.append((s, channels[key]))
file_path = 'nondeterminism_5.txt'
mode = RecordMode(file_path=file_path, replay=0)
f = theano.function([],
mode=mode,
updates=updates,
on_unused_input='ignore',
name='f')
for i in xrange(100):
disturb_mem.disturb_mem()
f()
mode.record.f.flush()
mode.record.f.close()
mode.set_record(Record(file_path=file_path, replay=1))
for i in xrange(100):
disturb_mem.disturb_mem()
f()
def run(replay):
disturb_mem.disturb_mem()
mode = RecordMode(file_path="nondeterminism_4.txt", replay=replay)
b = sharedX(np.zeros((1, )), name='b')
channels = OrderedDict()
disturb_mem.disturb_mem()
v_max = b.max(axis=0)
v_min = b.min(axis=0)
v_range = v_max - v_min
updates = []
for i, val in enumerate([
v_max.max(),
v_max.min(),
v_range.max(),
]):
disturb_mem.disturb_mem()
s = sharedX(0., name='s_' + str(i))
updates.append((s, val))
for var in theano.gof.graph.ancestors(update for var, update in updates):
if var.name is not None and var.name is not 'b':
if var.name[0] != 's' or len(var.name) != 2:
var.name = None
for key in channels:
updates.append((s, channels[key]))
f = theano.function([],
mode=mode,
updates=updates,
on_unused_input='ignore',
name='f')
for output in f.maker.fgraph.outputs:
mode.record.handle_line(var_descriptor(output) + '\n')
disturb_mem.disturb_mem()
f()
mode.record.f.flush()
mode.record.f.close()
def run(replay):
disturb_mem.disturb_mem()
mode = RecordMode(file_path= "nondeterminism_4.txt",
replay=replay)
b = sharedX(np.zeros((2,)), name='b')
channels = OrderedDict()
disturb_mem.disturb_mem()
v_max = b.max(axis=0)
v_min = b.min(axis=0)
v_range = v_max - v_min
updates = []
for i, val in enumerate([
v_max.max(),
v_max.min(),
v_range.max(),
]):
disturb_mem.disturb_mem()
s = sharedX(0., name='s_'+str(i))
updates.append((s, val))
for var in theano.gof.graph.ancestors(update for var, update in updates):
if var.name is not None and var.name is not 'b':
if var.name[0] != 's' or len(var.name) != 2:
var.name = None
for key in channels:
updates.append((s, channels[key]))
f = theano.function([], mode=mode, updates=updates, on_unused_input='ignore', name='f')
for output in f.maker.fgraph.outputs:
mode.record.handle_line(var_descriptor(output)+'\n')
disturb_mem.disturb_mem()
f()
mode.record.f.flush()
mode.record.f.close()
from pylearn2.config import yaml_parse
from pylearn2.devtools import disturb_mem
import sys
disturb_mem.disturb_mem()
_, replay = sys.argv
replay = int(replay)
disturb_mem.disturb_mem()
disturb_mem.disturb_mem()
disturb_mem.disturb_mem()
disturb_mem.disturb_mem()
disturb_mem.disturb_mem()
disturb_mem.disturb_mem()
yaml_src = """
!obj:pylearn2.train.Train {
dataset: &train !obj:pylearn2.datasets.mnist.MNIST {
which_set: "train",
#binarize: 1,
one_hot: 1,
start: 0,
stop: 100
},
model: !obj:galatea.dbm.inpaint.super_dbm.MLP_Wrapper {
decapitate: 0,
super_dbm: !obj:galatea.dbm.inpaint.super_dbm.SuperDBM {
batch_size : 100,
niter: 6, #note: since we have to backprop through the whole thing, this does
#increase the memory usage