class Trainer:
def __init__(self):
self.model = None
self.batch_size = 0
self.x_train = None
self.x_test = None
self.y_train = None
self.y_test = None
def train(self, epoch=10, batch_size=32, gpu=False):
if gpu:
cuda.check_cuda_available()
xp = cuda.cupy if gpu else np
self.batch_size = batch_size
label_types = ['none', 'tap', 'up', 'down', 'right', 'left']
self.model = Alex(len(label_types))
optimizer = optimizers.MomentumSGD(lr=0.01, momentum=0.9)
optimizer.setup(self.model)
if gpu:
self.model.to_gpu()
training_data = TrainingData(IMAGE_ROOT, NOTE_ROOT, VIDEO_ROOT, SONG_LIST_PATH)
self.x_train, self.x_test, self.y_train, self.y_test = training_data.get_train_data(label_types)
data_size = self.x_train.shape[0]
for ep in range(epoch):
print('epoch {0}/{1}: (learning rate={2})'.format(ep + 1, epoch, optimizer.lr))
indexes = np.random.permutation(data_size)
for i in range(0, data_size, self.batch_size):
x_batch = self.x_train[indexes[i:i + self.batch_size]]
y_batch = self.y_train[indexes[i:i + self.batch_size]]
x = chainer.Variable(x_batch)
t = chainer.Variable(y_batch)
optimizer.update(self.model, x, t)
print("loss: {0}".format(self.model.loss.data))
serializers.save_npz(MODEL_PATH, self.model)
optimizer.lr *= 0.97
def test(self):
print('Test started')
test_data_size = self.x_test.shape[0]
sum_loss = 0
for i in range(0, test_data_size, self.batch_size):
x = chainer.Variable(self.x_test[i:i + self.batch_size])
t = chainer.Variable(self.y_test[i:i + self.batch_size])
loss = self.model(x, t)
sum_loss += loss.data * self.batch_size
print(self.model.forward(x).data)
print('mean loss: {0}'.format(sum_loss / test_data_size))
self.root.fill('hy',y)
self.root.fill('hz',z)
self.root.fill('tgauss',tup)
return True
#------- execution
if __name__ == '__main__':
# imports
from alex import Alex
from messenger import Message, Messenger
from rootsvc import ROOTSvc
# configure alex
alex = Alex()
alex.nevts = 1000 # number of events to run
alex.evtfreq = 100 # frequency to print the event number
# create messenger service, log output in file 'alex.log', level INFO
msg = Messenger(name='msg',fname='alex_rootwrite.log',level=Message.Info)
alex.addsvc(msg)
# create histogram service, histogram root file 'alex.root'
root = ROOTSvc(name='root',fname='alex_rootwrite.root')
alex.addsvc(root)
#flow
gaus = GaussGenerate(name='gaus') # create GaussGenerate algorithm
alex.addalg(gaus) # add gaus algorithm to alex flow
def define(self):
""" path a list of the keys of the event DataSvc to dump
"""
self.paths = []
return
def execute(self):
for path in self.paths:
ok,data = self.evt.has(path),self.evt[path]
self.msg.info(self.name,path,ok)
if (ok): self.msg.info(self.name,data)
return True
# example of how to use the algorithms
if __name__ == '__main__':
from alex import Alex
alex = Alex('alex')
alex.nevts = 100
eff = EffFilter('eff')
eff.eff = 0.8
alex.addalg(eff)
alex.run()
""" finalize
"""
if (self.reader): self.reader.close()
return
#-------- how to use reader
if __name__ == "__main__":
# imports
from alex import Alex
from algorithms import Dump
# configure alex
alex = Alex('alex')
# flow
reader = Reader('reader',freader=FileReader)
reader.fnames = ['examples/data0.txt','examples/data1.txt',
'examples/data2.txt']
alex.addalg(reader)
dump = Dump('dump')
dump.paths+=['reader']
alex.addalg(dump)
# run
alex.run()
# for seg in segments:
#
# # Get the mean, min, and max chi2 values for each segment.
# # Set to -1 if the segment is not >= 3 points.
# mean_chi2 = -1.; min_chi2 = -1.; max_chi2 = -1.;
# if(len(seg.seg_k) >= 2):
# mean_chi2 = np.mean(seg.seg_fchisq[1:]);
# min_chi2 = min(seg.seg_fchisq[1:]);
# max_chi2 = max(seg.seg_fchisq[1:]);
#
# # Print the segment file line.
# f_fseg.write("{0} {1} {2} {3} {4}\n".format(seg.seg_id,len(seg.seg_k),mean_chi2,min_chi2,max_chi2));
#
# # Print the track file lines for each segment point: include the smeared hits.
# for k,x0,y0,z0,p1p,p2p,p3p,p4p,chi2p,p1f,p2f,p3f,p4f,chi2f,cfxy,cftxy,edep in zip(seg.seg_k,seg.seg_x0,seg.seg_y0,seg.seg_z0,seg.seg_p1p,seg.seg_p2p,seg.seg_p3p,seg.seg_p4p,seg.seg_pchisq,seg.seg_p1f,seg.seg_p2f,seg.seg_p3f,seg.seg_p4f,seg.seg_fchisq,seg.seg_cfxy,seg.seg_cftxy,seg.seg_edep):
# if(isinstance(chi2f,KFVector)):
# chi2 = chi2f[2];
# else:
# chi2 = chi2f;
# f_ftrk.write("{0} {1} {2} {3} {4} {5} {6} {7} {8} {9} {10} {11} {12} {13} {14} {15} {16} {17}\n".format(seg.seg_id,k,x0,y0,z0,p1p,p2p,p3p,p4p,chi2p,p1f,p2f,p3f,p4f,chi2f,cfxy,cftxy,edep));
# return True
def finalize(self):
return
alex = Alex('alex')
kf = JJKalmanFilter()
alex.addalg(kf)
alex.nevts = 1
alex.run()
# for seg in segments:
#
# # Get the mean, min, and max chi2 values for each segment.
# # Set to -1 if the segment is not >= 3 points.
# mean_chi2 = -1.; min_chi2 = -1.; max_chi2 = -1.;
# if(len(seg.seg_k) >= 2):
# mean_chi2 = np.mean(seg.seg_fchisq[1:]);
# min_chi2 = min(seg.seg_fchisq[1:]);
# max_chi2 = max(seg.seg_fchisq[1:]);
#
# # Print the segment file line.
# f_fseg.write("{0} {1} {2} {3} {4}\n".format(seg.seg_id,len(seg.seg_k),mean_chi2,min_chi2,max_chi2));
#
# # Print the track file lines for each segment point: include the smeared hits.
# for k,x0,y0,z0,p1p,p2p,p3p,p4p,chi2p,p1f,p2f,p3f,p4f,chi2f,cfxy,cftxy,edep in zip(seg.seg_k,seg.seg_x0,seg.seg_y0,seg.seg_z0,seg.seg_p1p,seg.seg_p2p,seg.seg_p3p,seg.seg_p4p,seg.seg_pchisq,seg.seg_p1f,seg.seg_p2f,seg.seg_p3f,seg.seg_p4f,seg.seg_fchisq,seg.seg_cfxy,seg.seg_cftxy,seg.seg_edep):
# if(isinstance(chi2f,KFVector)):
# chi2 = chi2f[2];
# else:
# chi2 = chi2f;
# f_ftrk.write("{0} {1} {2} {3} {4} {5} {6} {7} {8} {9} {10} {11} {12} {13} {14} {15} {16} {17}\n".format(seg.seg_id,k,x0,y0,z0,p1p,p2p,p3p,p4p,chi2p,p1f,p2f,p3f,p4f,chi2f,cfxy,cftxy,edep));
# return True
def finalize( self ):
return
alex = Alex('alex')
kf = JJKalmanFilter()
alex.addalg(kf)
alex.nevts = 1
alex.run()
