def main():
args = get_args()
train_data = None
train_label = []
for i in range(1, 6):
d = unpickle("cifar-10-batches-py/data_batch_%d" % i)
if i == 1:
train_data = d['data']
else:
train_data = np.vstack((train_data, d['data']))
train_label = train_label + d['labels']
train_data = train_data.reshape((len(train_data), 3, 32, 32))
train_label = np.array(train_label)
test_d = unpickle("cifar-10-batches-py/test_batch")
test_data = test_d['data']
test_data = test_data.reshape((len(test_data), 3, 32, 32))
test_label = np.array(test_d['labels'])
train_data = train_data.astype(np.float32)
test_data = test_data.astype(np.float32)
train_data /= 255
test_data /= 255
train_label = train_label.astype(np.int32)
test_label = test_label.astype(np.int32)
train = chainer.datasets.tuple_dataset.TupleDataset(
train_data, train_label)
test = chainer.datasets.tuple_dataset.TupleDataset(test_data, test_label)
#train, test = chainer.datasets.get_cifar10()
minibatch_size = args.batch_size
train_iter = iterators.SerialIterator(train, minibatch_size)
test_iter = iterators.SerialIterator(test, minibatch_size, False, False)
model = Model()
if args.gpu_idx >= 0:
chainer.cuda.get_device(args.gpu_idx).use()
model.to_gpu(args.gpu_idx)
model = L.Classifier(model)
optimizer = optimizers.Adam(eps=args.learning_rate)
optimizer.setup(model)
updater = training.updaters.StandardUpdater(train_iter,
optimizer,
device=args.gpu_idx)
max_epoch = args.epoch
trainer = training.Trainer(updater, (max_epoch, 'epoch'),
out='cifar10_result')
#save the log
trainer.extend(extensions.LogReport())
#save the model per 5 epoch
trainer.extend(extensions.snapshot(), trigger=(5, 'epoch'))
#validation
trainer.extend(extensions.Evaluator(test_iter, model, device=args.gpu_idx))
#print the log
trainer.extend(
extensions.PrintReport([
'epoch', 'main/loss', 'main/accuracy', 'validation/main/loss',
'validation/main/accuracy', 'elapsed_time'
]))
#save the glaph
trainer.extend(
extensions.dump_graph(root_name="main/loss", out_name="cg.dot"))
#save the loss glaph
trainer.extend(
extensions.PlotReport(['main/loss', 'validation/main/loss'],
x_key='epoch',
file_name='loss.png'))
#save the accuracy glaph
trainer.extend(
extensions.PlotReport(['main/accuracy', 'validation/main/accuracy'],
x_key='epoch',
file_name='accuracy.png'))
#print the progress bar
trainer.extend(extensions.ProgressBar())
trainer.run()
def main():
args = get_args()
train, test = chainer.datasets.get_mnist()
minibatch_size = args.batch_size
train_iter = iterators.SerialIterator(train, minibatch_size)
test_iter = iterators.SerialIterator(test, minibatch_size, False, False)
model = Model()
if args.gpu_idx >= 0:
chainer.cuda.get_device(args.gpu_idx).use()
model.to_gpu(args.gpu_idx)
model = L.Classifier(model)
optimizer = optimizers.Adam(eps=args.learning_rate)
optimizer.setup(model)
updater = training.updaters.StandardUpdater(train_iter,
optimizer,
device=args.gpu_idx)
max_epoch = args.epoch
trainer = training.Trainer(updater, (max_epoch, 'epoch'),
out='mnist_result')
#save the log
trainer.extend(extensions.LogReport())
#save the model per 5 epoch
trainer.extend(extensions.snapshot(), trigger=(5, 'epoch'))
#validation
trainer.extend(extensions.Evaluator(test_iter, model, device=args.gpu_idx))
#print the log
trainer.extend(
extensions.PrintReport([
'epoch', 'main/loss', 'main/accuracy', 'validation/main/loss',
'validation/main/accuracy', 'elapsed_time'
]))
#save the glaph
trainer.extend(
extensions.dump_graph(root_name="main/loss", out_name="cg.dot"))
#save the loss glaph
trainer.extend(
extensions.PlotReport(['main/loss', 'validation/main/loss'],
x_key='epoch',
file_name='loss.png'))
#save the accuracy glaph
trainer.extend(
extensions.PlotReport(['main/accuracy', 'validation/main/accuracy'],
x_key='epoch',
file_name='accuracy.png'))
#print the progress bar
trainer.extend(extensions.ProgressBar())
trainer.run()
