hidden1 = 300 #hidden layer 1
hidden2 = 100 #hidden layer 2
acti_type='tanh' #activation type
epoch = 100 #epochs number
advertiser = '2997'
if len(sys.argv) > 1:
advertiser = sys.argv[1]
train_file='../../make-ipinyou-data/' + advertiser + '/train.fm.txt' #training file
test_file='../../make-ipinyou-data/' + advertiser + '/test.fm.txt' #test file
fm_model_file='../../make-ipinyou-data/' + advertiser + '/fm.model.txt' #fm model file
#feats = ut.feats_len(train_file) #feature size
if len(sys.argv) > 2:
train_file=train_file+'.10.txt'
print train_file
train_size=ut.file_len(train_file) #training size
test_size=ut.file_len(test_file) #test size
n_batch=train_size/batch_size #number of batches
x_dim=133465
#best parameters
if advertiser=='2997': #best 0.615759999486
hidden0=200
hidden1=300
hidden2=100
lr=0.001
lambda1=0.0000001
dropout=0.98
lambda1=0
elif advertiser=='3386':
train_size=ut.file_len(train_file) #training size
acti_type='tanh' #activation type
epoch = 100 #epochs number
advertiser = '2997'
if len(sys.argv) > 1:
advertiser = sys.argv[1]
train_file='../data/train.fm.txt' #training file
test_file='../data/test.fm.txt' #test file
fm_model_file='../data/fm.model.txt' #fm model file
#feats = ut.feats_len(train_file) #feature size
if len(sys.argv) > 2 and advertiser=='all':
train_file=train_file+'.5.txt'
elif len(sys.argv) > 2:
train_file=train_file+'.10.txt'
print(train_file)
train_size=ut.file_len(train_file) #training size
test_size=ut.file_len(test_file) #test size
n_batch=train_size/batch_size #number of batches
x_drop=1
if advertiser=='2997':#
lr=0.001
x_drop=dropout=0.5
hidden1=300
hidden2=100
lambda1=0.0
lambda_fm=0.1
fi = open(train_file, 'r')
for line in fi:
if line.strip() != '':
s = line.strip().replace(':', ' ').split(' ')
fi=[]
for f in range(1, len(s), 2):
if int(s[f+1])==1:
fi.append(int(s[f]))
farr.append(fi)
yarr.append(int(s[0]))
train_size+=1
farr = numpy.array(farr, dtype = numpy.int32)
yarr = numpy.array(yarr, dtype = numpy.int32)
test_size=ut.file_len(test_file) #test size
n_batch=train_size/batch_size #number of batches
x_dim=133465
if advertiser == '2997':
lr=0.05
if advertiser== '3386': #number of batches
x_dim=0
if sys.argv[2]=='mod' and advertiser=='2997':
lr=0.1
lambda1=0.00
rng.seed(1234)
batch_size=10 #batch size
lr=0.01 # 0.002 #learning rate
lambda1=0.001 # 100 # 3 # .1 #regularisation rate
hidden1 = 300 #hidden layer 1
hidden2 = 100 #hidden layer 2
acti_type='tanh' #activation type
epoch = 100 #epochs number
advertiser = '2997'
if len(sys.argv) > 1:
advertiser = sys.argv[1]
train_file='../../make-ipinyou-data/' + advertiser + '/train.fm.txt' #training file
test_file='../../make-ipinyou-data/' + advertiser + '/test.fm.txt' #test file
fm_model_file='../../make-ipinyou-data/' + advertiser + '/fm.model.txt' #fm model file
#feats = ut.feats_len(train_file) #feature size
train_size=ut.file_len(train_file)
test_size=ut.file_len(test_file)
# train_size=312437 #ut.file_len(train_file) #training size
# test_size=156063 #ut.file_len(test_file) #test size
n_batch=train_size/batch_size #number of batches
name_field = {'weekday':0, 'hour':1, 'useragent':2, 'IP':3, 'region':4, 'city':5, 'adexchange':6, 'domain':7, 'slotid':8,
'slotwidth':9, 'slotheight':10, 'slotvisibility':11, 'slotformat':12, 'creative':13, 'advertiser':14, 'slotprice':15}
feat_field = {}
feat_weights = {}
w_0 = 0
feat_num = 0
k = 0
