def CifarAnalysis(folderName=None, batchsize=1000, **kwd):
id_gpu = 0
OutStr = ""
OutStr += 'GPU: {}\n'.format(id_gpu)
OutStr += 'Minibatch-size: {}\n'.format(batchsize)
OutStr += 'kwd: {}\n'.format(kwd)
OutStr += ''
print OutStr
fOutput = None
if folderName:
if not os.path.exists(folderName):
os.makedirs(folderName)
fOutput = open(os.path.join(folderName, "output.dat"), "w")
shutil.copyfile(__file__,
os.path.join(folderName, os.path.basename(__file__)))
# Prepare dataset
data_tr = np.zeros((100000, 3 * 32 * 32), dtype=np.float32)
data_ev = np.zeros((10000, 3 * 32 * 32), dtype=np.float32)
label_tr = np.zeros((100000), dtype=np.int32)
label_ev = np.zeros((10000), dtype=np.int32)
I_colors = 3
I_Xunit = 32
I_Yunit = 32
F_unit = 100 # be careful!!
h5f_tr = h5py.File("data_cifar100/train.h5f", "r")
data_tr[:50000] = h5f_tr["ZCA_byTrainData/data"].value
label_tr[:50000] = h5f_tr["Info/fine_labels"].value
data_tr[50000:] = h5f_tr["ZCA_byTrainData/data"].value
label_tr[50000:] = h5f_tr["Info/fine_labels"].value
x_tr = data_tr.reshape((len(data_tr), 3, 32, 32))
x_tr[50000:] = x_tr[:50000][:, :, :, ::-1]
#x1_tr = x_tr.copy()
#x1_tr = (x_tr[:50000])
#x2_tr = x_tr.copy()
#x2_tr = (x_tr[:50000])[:,:,:,::-1]
#x_tr[:50000] = (x_tr[:50000])[:,:,:,::-1]
#plt.subplot(211)
#plt.imshow(x_tr[1].transpose(1,2,0))
#plt.subplot(212)
#plt.imshow(x_tr[50001].transpose(1,2,0))
#plt.show()
#sys.exit(-1)
h5f_ev = h5py.File("data_cifar100/test.h5f", "r")
data_ev[:] = h5f_ev["ZCA_byTrainData/data"].value
label_ev[:] = h5f_ev["Info/fine_labels"].value
## Prep
x_ev = data_ev.reshape((len(data_ev), 3, 32, 32))
y_tr = label_tr
y_ev = label_ev
N_tr = len(data_tr) # 50000
N_ev = len(data_ev) # 10000
## Define analisis
Resume = None
if "Resume" in kwd:
Resume = kwd["Resume"]
del kwd["Resume"]
model, ModelKwd = net.GenModel(I_colors=I_colors,
I_Xunit=I_Xunit,
I_Yunit=I_Yunit,
F_unit=F_unit,
**kwd)
if id_gpu >= 0:
cuda.get_device(id_gpu).use()
model.to_gpu()
xp = np if id_gpu < 0 else cuda.cupy
# Setup optimizer
optimizer = optimizers.Adam()
optimizer.setup(model)
# Init/Resume
if Resume:
print 'Load optimizer state from %s' % (Resume)
with h5py.File(Resume, "r") as f:
s = HDF5Deserializer(f)
s_model = s["model"]
s_model.load(model)
# Setup stop manager
sm = StopManager.StopManager()
sm.SetMaximumEpoch(10000)
sm.SetMinimumEpoch(10)
sm.SetStopThreshold(3e-4)
print sm
# Learning loop
if fOutput: fOutput.write("epoch,mode,loss,accuracy\n")
#for epoch in six.moves.range(1, n_epoch + 1):
epoch = 0
while True:
epoch += 1
print 'epoch %d' % epoch
# training
perm = np.random.permutation(N_tr)
sum_accuracy = 0
sum_loss = 0
start = time.time()
for i in six.moves.range(0, N_tr, batchsize):
x = chainer.Variable(xp.asarray(x_tr[perm[i:i + batchsize]]))
t = chainer.Variable(xp.asarray(y_tr[perm[i:i + batchsize]]))
# Pass the loss function (Classifier defines it) and its arguments
model.predictor.setTrainMode(True)
optimizer.update(model, x, t)
if (epoch == 1 and i == 0) and folderName:
with open(os.path.join(folderName, 'graph.dot'), 'w') as o:
g = computational_graph.build_computational_graph(
(model.loss, ))
o.write(g.dump())
print 'graph generated'
sum_loss += float(model.loss.data) * len(t.data)
sum_accuracy += float(model.accuracy.data) * len(t.data)
end = time.time()
elapsed_time = end - start
throughput = N_tr / elapsed_time
print 'train mean loss=%.5f, accuracy=%.2f%%, throughput=%.0f images/sec' % (
sum_loss / N_tr, sum_accuracy / N_tr * 100., throughput)
if fOutput:
fOutput.write("%d,Train,%e,%e\n" %
(epoch, sum_loss / N_tr, sum_accuracy / N_tr))
# evaluation
perm = np.random.permutation(N_ev)
sum_accuracy = 0
sum_loss = 0
for i in six.moves.range(0, N_ev, batchsize):
x = chainer.Variable(xp.asarray(x_ev[perm[i:i + batchsize]]),
volatile='on')
t = chainer.Variable(xp.asarray(y_ev[perm[i:i + batchsize]]),
volatile='on')
model.predictor.setTrainMode(False)
loss = model(x, t)
sum_loss += float(loss.data) * len(t.data)
sum_accuracy += float(model.accuracy.data) * len(t.data)
print 'test mean loss=%.5f, accuracy=%.2f%%' % (
sum_loss / N_ev,
sum_accuracy / N_ev * 100,
)
sm.AddAccuracy(sum_accuracy / N_ev)
print sm.GetInfo()
if fOutput:
fOutput.write("%d,Test,%e,%e\n" %
(epoch, sum_loss / N_ev, sum_accuracy / N_ev))
StopFlag = sm.StopCheck()
if folderName and (epoch % 1 == 0 or StopFlag):
# Save the model and the optimizer
if StopFlag:
myFname = os.path.join(folderName, 'mlp_final')
else:
myFname = os.path.join(folderName, 'mlp_%d' % epoch)
with h5py.File(myFname + ".hdf5", "w") as f:
s = HDF5Serializer(f)
s["model"].save(model)
f.create_dataset("kwd",
data=ModelKwd.__str__(),
dtype=h5py.special_dtype(vlen=unicode))
f.create_dataset("net",
data=netFile,
dtype=h5py.special_dtype(vlen=unicode))
f.flush()
if StopFlag: break
if fOutput: fOutput.close()
def CifarAnalysis(folderName=None, n_epoch=8, batchsize=1000, **kwd):
id_gpu = 0
OutStr = ""
OutStr += 'GPU: {}\n'.format(id_gpu)
OutStr += 'Minibatch-size: {}\n'.format(batchsize)
OutStr += 'kwd: {}\n'.format(kwd)
OutStr += ''
print OutStr
fOutput = None
if folderName:
if not os.path.exists(folderName):
os.makedirs(folderName)
fOutput = open(os.path.join(folderName, "output.dat"), "w")
shutil.copyfile(__file__,
os.path.join(folderName, os.path.basename(__file__)))
# Prepare dataset
h5f = h5py.File("input.dat", "r")
data = h5f["Original/Data"].value
label = h5f["Info/Label"].value
imIndex = h5f["Info/ImIndex"].value
agIndex = h5f["Info/AgIndex"].value
TraEva = np.random.random(max(imIndex) + 1).repeat(3)
TraEva = TraEva < 0.8
with h5py.File(os.path.join(folderName, "TraEva.hdf5"), "w") as af:
af.create_dataset("TraEva", data=TraEva, dtype=TraEva.dtype)
af.flush()
data_tr = data[TraEva]
label_tr = label[TraEva]
data_ev = data[np.logical_not(TraEva)]
label_ev = label[np.logical_not(TraEva)]
agIndex_ev = agIndex[np.logical_not(TraEva)]
data_ev = data_ev[agIndex_ev == 0]
label_ev = label_ev[agIndex_ev == 0]
I_colors = 3
I_Xunit = 32
I_Yunit = 32
F_unit = max(label) + 1 # be careful!!
#print np.mean(data_tr,axis=1).reshape(3*32*32)
data_tr -= np.mean(data_tr, axis=1).reshape((len(data_tr), 1))
data_tr -= np.mean(data_tr, axis=0)
data_tr /= np.std(data_tr, axis=0)
data_ev -= np.mean(data_ev, axis=1).reshape((len(data_ev), 1))
data_ev -= np.mean(data_ev, axis=0)
data_ev /= np.std(data_ev, axis=0)
## Prep
x_tr = data_tr.reshape((len(data_tr), 3, 32, 32))
x_ev = data_ev.reshape((len(data_ev), 3, 32, 32))
y_tr = label_tr
y_ev = label_ev
N_tr = len(data_tr) # 50000
N_ev = len(data_ev) # 10000
## Define analisis
Resume = None
if "Resume" in kwd:
Resume = kwd["Resume"]
del kwd["Resume"]
#model,ModelKwd = net2.GenModel(F_unit = F_unit)
model = net.GenModel(F_unit)
ModelKwd = ""
if id_gpu >= 0:
cuda.get_device(id_gpu).use()
model.to_gpu()
xp = np if id_gpu < 0 else cuda.cupy
# Setup optimizer
optimizer = optimizers.Adam()
optimizer.setup(model)
#optimizer.add_hook(scheduled_alpha_reduction)
# Init/Resume
if Resume:
print 'Load optimizer state from %s' % (Resume)
with h5py.File(Resume, "r") as f:
s = HDF5Deserializer(f)
s_model = s["model"]
s_model.load(model)
# Setup stop manager
sm = StopManager.StopManager()
sm.SetMaximumEpoch(10000)
sm.SetMinimumEpoch(10)
sm.SetStopThreshold(3e-4)
print sm
#alphaTiming = [10,20,40,80]
#optimizer.alpha /= 16
# Learning loop
if fOutput: fOutput.write("epoch,mode,loss,accuracy\n")
#for epoch in six.moves.range(1, n_epoch + 1):
epoch = 0
while True:
epoch += 1
print 'epoch %d' % epoch
# training
perm = np.random.permutation(N_tr)
sum_accuracy = 0
sum_loss = 0
start = time.time()
"""
if epoch in alphaTiming:
optimizer.alpha /= 2
print "alpha changed... currently Alpha = ", optimizer.alpha
"""
for i in six.moves.range(0, N_tr, batchsize):
bx = x_tr[perm[i:i + batchsize]]
#bx = ag.Aug(bx)
x = chainer.Variable(xp.asarray(bx))
t = chainer.Variable(xp.asarray(y_tr[perm[i:i + batchsize]]))
# Pass the loss function (Classifier defines it) and its arguments
model.predictor.setTrainMode(True)
optimizer.update(model, x, t)
if (epoch == 1 and i == 0) and folderName:
with open(os.path.join(folderName, 'graph.dot'), 'w') as o:
g = computational_graph.build_computational_graph(
(model.loss, ))
o.write(g.dump())
print 'graph generated'
sum_loss += float(model.loss.data) * len(t.data)
sum_accuracy += float(model.accuracy.data) * len(t.data)
end = time.time()
elapsed_time = end - start
throughput = N_tr / elapsed_time
print 'train mean loss=%.5f, accuracy=%.2f%%, throughput=%.0f images/sec' % (
sum_loss / N_tr, sum_accuracy / N_tr * 100., throughput)
if fOutput:
fOutput.write("%d,Train,%e,%e\n" %
(epoch, sum_loss / N_tr, sum_accuracy / N_tr))
# evaluation
perm = np.random.permutation(N_ev)
sum_accuracy = 0
sum_loss = 0
for i in six.moves.range(0, N_ev, batchsize):
x = chainer.Variable(xp.asarray(x_ev[perm[i:i + batchsize]]),
volatile='on')
t = chainer.Variable(xp.asarray(y_ev[perm[i:i + batchsize]]),
volatile='on')
model.predictor.setTrainMode(False)
loss = model(x, t)
sum_loss += float(loss.data) * len(t.data)
sum_accuracy += float(model.accuracy.data) * len(t.data)
print 'test mean loss=%.5f, accuracy=%.2f%%' % (
sum_loss / N_ev,
sum_accuracy / N_ev * 100,
)
sm.AddAccuracy(sum_accuracy / N_ev)
print sm.GetInfo()
if fOutput:
fOutput.write("%d,Test,%e,%e\n" %
(epoch, sum_loss / N_ev, sum_accuracy / N_ev))
StopFlag = sm.StopCheck()
if epoch >= n_epoch: StopFlag = True
#StopFlag = False
if folderName and (epoch % 1 == 0 or StopFlag):
# Save the model and the optimizer
if StopFlag:
myFname = os.path.join(folderName, 'mlp_final')
else:
myFname = os.path.join(folderName, 'mlp_%d' % epoch)
with h5py.File(myFname + ".hdf5", "w") as f:
s = HDF5Serializer(f)
s["model"].save(model)
f.create_dataset("kwd",
data=ModelKwd.__str__(),
dtype=h5py.special_dtype(vlen=unicode))
f.create_dataset("net",
data=netFile,
dtype=h5py.special_dtype(vlen=unicode))
f.flush()
if StopFlag: break
if fOutput: fOutput.close()