if len(args) > 0 and args[0] == 'sigmoid':
type = 0
elif len(args) > 0 and args[0] == 'tanh':
type = 1
part = 2 #0=train, 1=valid, 2=test
PATH = '' #Can be changed too if model is not in the current drectory
if os.path.exists(PATH+'params_finetune_NIST.txt'):
start_time = time.clock()
print ('\n finetune = NIST ')
print "NIST DIGITS"
test_data(PATH+'params_finetune_NIST.txt',datasets.nist_digits(),part=part,type=type)
print "NIST LOWER CASE"
test_data(PATH+'params_finetune_NIST.txt',datasets.nist_lower(),part=part,type=type)
print "NIST UPPER CASE"
test_data(PATH+'params_finetune_NIST.txt',datasets.nist_upper(),part=part,type=type)
end_time = time.clock()
print ('It took %f minutes' %((end_time-start_time)/60.))
if os.path.exists(PATH+'params_finetune_P07.txt'):
start_time = time.clock()
print ('\n finetune = P07 ')
print "NIST DIGITS"
test_data(PATH+'params_finetune_P07.txt',datasets.nist_digits(),part=part,type=type)
print "NIST LOWER CASE"
test_data(PATH+'params_finetune_P07.txt',datasets.nist_lower(),part=part,type=type)
print "NIST UPPER CASE"
test_data(PATH+'params_finetune_P07.txt',datasets.nist_upper(),part=part,type=type)
def jobman_entrypoint(state, channel):
# record mercurial versions of each package
pylearn.version.record_versions(state,[theano,ift6266,pylearn])
# TODO: remove this, bad for number of simultaneous requests on DB
channel.save()
# For test runs, we don't want to use the whole dataset so
# reduce it to fewer elements if asked to.
rtt = None
if state.has_key('reduce_train_to'):
rtt = state['reduce_train_to']
elif REDUCE_TRAIN_TO:
rtt = REDUCE_TRAIN_TO
n_ins = 32*32
n_outs = 62 # 10 digits, 26*2 (lower, capitals)
examples_per_epoch = NIST_ALL_TRAIN_SIZE
PATH = ''
NIST_BY_CLASS=0
print "Creating optimizer with state, ", state
optimizer = SdaSgdOptimizer(dataset=datasets.nist_all(),
hyperparameters=state, \
n_ins=n_ins, n_outs=n_outs,\
examples_per_epoch=examples_per_epoch, \
max_minibatches=rtt)
if os.path.exists(PATH+'params_finetune_NIST.txt'):
print ('\n finetune = NIST ')
optimizer.reload_parameters(PATH+'params_finetune_NIST.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
if os.path.exists(PATH+'params_finetune_P07.txt'):
print ('\n finetune = P07 ')
optimizer.reload_parameters(PATH+'params_finetune_P07.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
if os.path.exists(PATH+'params_finetune_NIST_then_P07.txt'):
print ('\n finetune = NIST then P07')
optimizer.reload_parameters(PATH+'params_finetune_NIST_then_P07.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
if os.path.exists(PATH+'params_finetune_P07_then_NIST.txt'):
print ('\n finetune = P07 then NIST')
optimizer.reload_parameters(PATH+'params_finetune_P07_then_NIST.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
if os.path.exists(PATH+'params_finetune_PNIST07.txt'):
print ('\n finetune = PNIST07')
optimizer.reload_parameters(PATH+'params_finetune_PNIST07.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
if os.path.exists(PATH+'params_finetune_PNIST07_then_NIST.txt'):
print ('\n finetune = PNIST07 then NIST')
optimizer.reload_parameters(PATH+'params_finetune_PNIST07_then_NIST.txt')
if NIST_BY_CLASS == 1:
print "NIST DIGITS"
optimizer.training_error(datasets.nist_digits(),part=2)
print "NIST LOWER CASE"
optimizer.training_error(datasets.nist_lower(),part=2)
print "NIST UPPER CASE"
optimizer.training_error(datasets.nist_upper(),part=2)
else:
print "P07 valid"
optimizer.training_error(datasets.nist_P07(),part=1)
print "PNIST valid"
optimizer.training_error(datasets.PNIST07(),part=1)
channel.save()
return channel.COMPLETE
def jobman_entrypoint(state, channel):
# record mercurial versions of each package
pylearn.version.record_versions(state,[theano,ift6266,pylearn])
# TODO: remove this, bad for number of simultaneous requests on DB
channel.save()
# For test runs, we don't want to use the whole dataset so
# reduce it to fewer elements if asked to.
rtt = None
if state.has_key('reduce_train_to'):
rtt = state['reduce_train_to']
elif REDUCE_TRAIN_TO:
rtt = REDUCE_TRAIN_TO
if state.has_key('decrease_lr'):
decrease_lr = state['decrease_lr']
else :
decrease_lr = 0
if state.has_key('decrease_lr_pretrain'):
dec=state['decrease_lr_pretrain']
else :
dec=0
n_ins = 32*32
if state.has_key('subdataset'):
subdataset_name=state['subdataset']
else:
subdataset_name=SUBDATASET_NIST
#n_outs = 62 # 10 digits, 26*2 (lower, capitals)
if subdataset_name == "upper":
n_outs = 26
subdataset = datasets.nist_upper()
examples_per_epoch = NIST_UPPER_TRAIN_SIZE
elif subdataset_name == "lower":
n_outs = 26
subdataset = datasets.nist_lower()
examples_per_epoch = NIST_LOWER_TRAIN_SIZE
elif subdataset_name == "digits":
n_outs = 10
subdataset = datasets.nist_digits()
examples_per_epoch = NIST_DIGITS_TRAIN_SIZE
else:
n_outs = 62
subdataset = datasets.nist_all()
examples_per_epoch = NIST_ALL_TRAIN_SIZE
print 'Using subdataset ', subdataset_name
#To be sure variables will not be only in the if statement
PATH = ''
nom_reptrain = ''
nom_serie = ""
if state['pretrain_choice'] == 0:
nom_serie="series_NIST.h5"
elif state['pretrain_choice'] == 1:
nom_serie="series_P07.h5"
series = create_series(state.num_hidden_layers,nom_serie)
print "Creating optimizer with state, ", state
optimizer = SdaSgdOptimizer(dataset_name=subdataset_name,\
dataset=subdataset,\
hyperparameters=state, \
n_ins=n_ins, n_outs=n_outs,\
examples_per_epoch=examples_per_epoch, \
series=series,
max_minibatches=rtt)
parameters=[]
#Number of files of P07 used for pretraining
nb_file=0
print('\n\tpretraining with NIST\n')
optimizer.pretrain(subdataset, decrease = dec)
channel.save()
#Set some of the parameters used for the finetuning
if state.has_key('finetune_set'):
finetune_choice=state['finetune_set']
else:
finetune_choice=FINETUNE_SET
if state.has_key('max_finetuning_epochs'):
max_finetune_epoch_NIST=state['max_finetuning_epochs']
else:
max_finetune_epoch_NIST=MAX_FINETUNING_EPOCHS
if state.has_key('max_finetuning_epochs_P07'):
max_finetune_epoch_P07=state['max_finetuning_epochs_P07']
else:
max_finetune_epoch_P07=max_finetune_epoch_NIST
#Decide how the finetune is done
if finetune_choice == 0:
print('\n\n\tfinetune with NIST\n\n')
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(subdataset,subdataset,max_finetune_epoch_NIST,ind_test=1,decrease=decrease_lr)
channel.save()
if finetune_choice == 1:
print('\n\n\tfinetune with P07\n\n')
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_P07(),datasets.nist_all(),max_finetune_epoch_P07,ind_test=0,decrease=decrease_lr)
channel.save()
if finetune_choice == 2:
print('\n\n\tfinetune with P07 followed by NIST\n\n')
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_P07(),datasets.nist_all(),max_finetune_epoch_P07,ind_test=20,decrease=decrease_lr)
optimizer.finetune(datasets.nist_all(),datasets.nist_P07(),max_finetune_epoch_NIST,ind_test=21,decrease=decrease_lr)
channel.save()
if finetune_choice == 3:
print('\n\n\tfinetune with NIST only on the logistic regression on top (but validation on P07).\n\
All hidden units output are input of the logistic regression\n\n')
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_all(),datasets.nist_P07(),max_finetune_epoch_NIST,ind_test=1,special=1,decrease=decrease_lr)
if finetune_choice==-1:
print('\nSERIE OF 4 DIFFERENT FINETUNINGS')
print('\n\n\tfinetune with NIST\n\n')
sys.stdout.flush()
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_all(),datasets.nist_P07(),max_finetune_epoch_NIST,ind_test=1,decrease=decrease_lr)
channel.save()
print('\n\n\tfinetune with P07\n\n')
sys.stdout.flush()
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_P07(),datasets.nist_all(),max_finetune_epoch_P07,ind_test=0,decrease=decrease_lr)
channel.save()
print('\n\n\tfinetune with P07 (done earlier) followed by NIST (written here)\n\n')
sys.stdout.flush()
optimizer.reload_parameters('params_finetune_P07.txt')
optimizer.finetune(datasets.nist_all(),datasets.nist_P07(),max_finetune_epoch_NIST,ind_test=21,decrease=decrease_lr)
channel.save()
print('\n\n\tfinetune with NIST only on the logistic regression on top.\n\
All hidden units output are input of the logistic regression\n\n')
sys.stdout.flush()
optimizer.reload_parameters('params_pretrain.txt')
optimizer.finetune(datasets.nist_all(),datasets.nist_P07(),max_finetune_epoch_NIST,ind_test=1,special=1,decrease=decrease_lr)
channel.save()
channel.save()
return channel.COMPLETE
def test_error(model_file):
print((' test error on all NIST'))
# load the model
a=numpy.load(model_file)
W1=a['W1']
W2=a['W2']
b1=a['b1']
b2=a['b2']
configuration=a['config']
#configuration = [learning_rate,nb_max_exemples,nb_hidden,adaptive_lr]
learning_rate = configuration[0]
nb_max_exemples = configuration[1]
nb_hidden = configuration[2]
adaptive_lr = configuration[3]
if(len(configuration) == 6):
detection_mode = configuration[4]
reduce_label = configuration[5]
else:
detection_mode = 0
reduce_label = 0
# define the batch size
batch_size=20
#define the nb of target
nb_targets = 62
# create the mlp
ishape = (32,32) # this is the size of NIST images
# allocate symbolic variables for the data
x = T.fmatrix() # the data is presented as rasterized images
y = T.lvector() # the labels are presented as 1D vector of
# [long int] labels
# construct the logistic regression class
classifier = MLP( input=x,\
n_in=32*32,\
n_hidden=nb_hidden,\
n_out=nb_targets,
learning_rate=learning_rate,\
detection_mode=detection_mode)
# set the weight into the model
classifier.W1.value = W1
classifier.b1.value = b1
classifier.W2.value = W2
classifier.b2.value = b2
# compiling a theano function that computes the mistakes that are made by
# the model on a minibatch
test_model = theano.function([x,y], classifier.errors(y))
# test it on the test set
# load NIST ALL
dataset=datasets.nist_all()
test_score = 0.
temp =0
for xt,yt in dataset.test(batch_size):
if reduce_label:
yt[yt > 35] = yt[yt > 35]-26
test_score += test_model(xt,yt)
temp = temp+1
test_score /= temp
print(( ' test error NIST ALL : %f %%') %(test_score*100.0))
# load NIST DIGITS
dataset=datasets.nist_digits()
test_score = 0.
temp =0
for xt,yt in dataset.test(batch_size):
if reduce_label:
yt[yt > 35] = yt[yt > 35]-26
test_score += test_model(xt,yt)
temp = temp+1
test_score /= temp
print(( ' test error NIST digits : %f %%') %(test_score*100.0))
# load NIST lower
dataset=datasets.nist_lower()
test_score = 0.
temp =0
for xt,yt in dataset.test(batch_size):
if reduce_label:
yt[yt > 35] = yt[yt > 35]-26
test_score += test_model(xt,yt)
temp = temp+1
test_score /= temp
print(( ' test error NIST lower : %f %%') %(test_score*100.0))
# load NIST upper
dataset=datasets.nist_upper()
test_score = 0.
temp =0
for xt,yt in dataset.test(batch_size):
if reduce_label:
yt[yt > 35] = yt[yt > 35]-26
test_score += test_model(xt,yt)
temp = temp+1
test_score /= temp
print(( ' test error NIST upper : %f %%') %(test_score*100.0))