def main():
archs = {
'alex': alex.Alex,
'alex_fp16': alex.AlexFp16,
'googlenet': googlenet.GoogLeNet,
'googlenetbn': googlenetbn.GoogLeNetBN,
'googlenetbn_fp16': googlenetbn.GoogLeNetBNFp16,
'nin': nin.NIN,
'resnet50': resnet50.ResNet50,
'resnext50': resnext50.ResNeXt50,
}
parser = argparse.ArgumentParser(
description='Learning convnet from ILSVRC2012 dataset')
parser.add_argument('train', help='Path to training image-label list file')
parser.add_argument('val', help='Path to validation image-label list file')
parser.add_argument('--arch', '-a', choices=archs.keys(), default='nin',
help='Convnet architecture')
parser.add_argument('--batchsize', '-B', type=int, default=32,
help='Learning minibatch size')
parser.add_argument('--epoch', '-E', type=int, default=10,
help='Number of epochs to train')
parser.add_argument('--device', '-d', type=str, default='-1',
help='Device specifier. Either ChainerX device '
'specifier or an integer. If non-negative integer, '
'CuPy arrays with specified device id are used. If '
'negative integer, NumPy arrays are used')
parser.add_argument('--initmodel',
help='Initialize the model from given file')
parser.add_argument('--loaderjob', '-j', type=int,
help='Number of parallel data loading processes')
parser.add_argument('--mean', '-m', default='mean.npy',
help='Mean file (computed by compute_mean.py)')
parser.add_argument('--resume', '-r', default='',
help='Initialize the trainer from given file')
parser.add_argument('--out', '-o', default='result',
help='Output directory')
parser.add_argument('--root', '-R', default='.',
help='Root directory path of image files')
parser.add_argument('--val_batchsize', '-b', type=int, default=250,
help='Validation minibatch size')
parser.add_argument('--test', action='store_true')
parser.set_defaults(test=False)
parser.add_argument('--dali', action='store_true')
parser.set_defaults(dali=False)
group = parser.add_argument_group('deprecated arguments')
group.add_argument('--gpu', '-g', type=int, nargs='?', const=0,
help='GPU ID (negative value indicates CPU)')
args = parser.parse_args()
device = parse_device(args)
print('Device: {}'.format(device))
print('# Minibatch-size: {}'.format(args.batchsize))
print('# epoch: {}'.format(args.epoch))
print('')
# Initialize the model to train
model = archs[args.arch]()
if args.initmodel:
print('Load model from {}'.format(args.initmodel))
chainer.serializers.load_npz(args.initmodel, model)
model.to_device(device)
device.use()
# Load the mean file
mean = np.load(args.mean)
if args.dali:
if not dali_util._dali_available:
raise RuntimeError('DALI seems not available on your system.')
num_threads = args.loaderjob
if num_threads is None or num_threads <= 0:
num_threads = 1
ch_mean = list(np.average(mean, axis=(1, 2)))
ch_std = [255.0, 255.0, 255.0]
# Setup DALI pipelines
train_pipe = dali_util.DaliPipelineTrain(
args.train, args.root, model.insize, args.batchsize,
num_threads, args.gpu, True, mean=ch_mean, std=ch_std)
val_pipe = dali_util.DaliPipelineVal(
args.val, args.root, model.insize, args.val_batchsize,
num_threads, args.gpu, False, mean=ch_mean, std=ch_std)
train_iter = chainer.iterators.DaliIterator(train_pipe)
val_iter = chainer.iterators.DaliIterator(val_pipe, repeat=False)
# converter = dali_converter
converter = dali_util.DaliConverter(mean=mean, crop_size=model.insize)
else:
# Load the dataset files
train = PreprocessedDataset(args.train, args.root, mean, model.insize)
val = PreprocessedDataset(args.val, args.root, mean, model.insize,
False)
# These iterators load the images with subprocesses running in parallel
# to the training/validation.
train_iter = chainer.iterators.MultiprocessIterator(
train, args.batchsize, n_processes=args.loaderjob)
val_iter = chainer.iterators.MultiprocessIterator(
val, args.val_batchsize, repeat=False, n_processes=args.loaderjob)
converter = dataset.concat_examples
# Set up an optimizer
optimizer = chainer.optimizers.MomentumSGD(lr=0.01, momentum=0.9)
optimizer.setup(model)
# Set up a trainer
updater = training.updaters.StandardUpdater(
train_iter, optimizer, converter=converter, device=device)
trainer = training.Trainer(updater, (args.epoch, 'epoch'), args.out)
val_interval = (1 if args.test else 100000), 'iteration'
log_interval = (1 if args.test else 1000), 'iteration'
trainer.extend(extensions.Evaluator(val_iter, model, converter=converter,
device=device), trigger=val_interval)
# TODO(sonots): Temporarily disabled for chainerx. Fix it.
if not (chainerx.is_available() and isinstance(device, chainerx.Device)):
trainer.extend(extensions.dump_graph('main/loss'))
trainer.extend(extensions.snapshot(), trigger=val_interval)
trainer.extend(extensions.snapshot_object(
model, 'model_iter_{.updater.iteration}'), trigger=val_interval)
# Be careful to pass the interval directly to LogReport
# (it determines when to emit log rather than when to read observations)
trainer.extend(extensions.LogReport(trigger=log_interval))
trainer.extend(extensions.observe_lr(), trigger=log_interval)
trainer.extend(extensions.PrintReport([
'epoch', 'iteration', 'main/loss', 'validation/main/loss',
'main/accuracy', 'validation/main/accuracy', 'lr'
]), trigger=log_interval)
trainer.extend(extensions.ProgressBar(update_interval=10))
if args.resume:
chainer.serializers.load_npz(args.resume, trainer)
trainer.run()
def main():
archs = {
'alex': alex.Alex,
'alex_fp16': alex.AlexFp16,
'googlenet': googlenet.GoogLeNet,
'googlenetbn': googlenetbn.GoogLeNetBN,
'googlenetbn_fp16': googlenetbn.GoogLeNetBNFp16,
'nin': nin.NIN,
'resnet50': resnet50.ResNet50,
'resnext50': resnext50.ResNeXt50,
}
parser = argparse.ArgumentParser(
description='Learning convnet from ILSVRC2012 dataset')
parser.add_argument('train', help='Path to training image-label list file')
parser.add_argument('val', help='Path to validation image-label list file')
parser.add_argument('--arch',
'-a',
choices=archs.keys(),
default='nin',
help='Convnet architecture')
parser.add_argument('--batchsize',
'-B',
type=int,
default=32,
help='Learning minibatch size')
parser.add_argument('--epoch',
'-E',
type=int,
default=10,
help='Number of epochs to train')
parser.add_argument('--gpu',
'-g',
type=int,
default=-1,
help='GPU ID (negative value indicates CPU')
parser.add_argument('--initmodel',
help='Initialize the model from given file')
parser.add_argument('--loaderjob',
'-j',
type=int,
help='Number of parallel data loading processes')
parser.add_argument('--mean',
'-m',
default='mean.npy',
help='Mean file (computed by compute_mean.py)')
parser.add_argument('--resume',
'-r',
default='',
help='Initialize the trainer from given file')
parser.add_argument('--out',
'-o',
default='result',
help='Output directory')
parser.add_argument('--root',
'-R',
default='.',
help='Root directory path of image files')
parser.add_argument('--val_batchsize',
'-b',
type=int,
default=250,
help='Validation minibatch size')
parser.add_argument('--test', action='store_true')
parser.set_defaults(test=False)
parser.add_argument('--dali', action='store_true')
parser.set_defaults(dali=False)
parser.add_argument(
'--nvtx_mark',
action='store_true',
help='Enable NVTX\'s marks during profiling by nvprof.')
parser.add_argument(
'--iter',
type=int,
default=0,
help=('Number of iterations for profiling.'
' NOTE: this option is preferred over epoch option.'))
args = parser.parse_args()
# Initialize the model to train
model = archs[args.arch]()
if args.initmodel:
print('Load model from {}'.format(args.initmodel))
chainer.serializers.load_npz(args.initmodel, model)
if args.gpu >= 0:
chainer.backends.cuda.get_device_from_id(
args.gpu).use() # Make the GPU current
model.to_gpu()
# Load the mean file
mean = np.load(args.mean)
if args.dali:
if not dali_util._dali_available:
raise RuntimeError('DALI seems not available on your system.')
num_threads = args.loaderjob
if num_threads is None or num_threads <= 0:
num_threads = 1
ch_mean = list(np.average(mean, axis=(1, 2)))
ch_std = [255.0, 255.0, 255.0]
# Setup DALI pipelines
train_pipe = dali_util.DaliPipelineTrain(args.train,
args.root,
model.insize,
args.batchsize,
num_threads,
args.gpu,
True,
mean=ch_mean,
std=ch_std)
val_pipe = dali_util.DaliPipelineVal(args.val,
args.root,
model.insize,
args.val_batchsize,
num_threads,
args.gpu,
False,
mean=ch_mean,
std=ch_std)
train_iter = chainer.iterators.DaliIterator(train_pipe)
val_iter = chainer.iterators.DaliIterator(val_pipe, repeat=False)
# converter = dali_converter
converter = dali_util.DaliConverter(mean=mean, crop_size=model.insize)
else:
# Load the dataset files
train = PreprocessedDataset(args.train, args.root, mean, model.insize)
val = PreprocessedDataset(args.val, args.root, mean, model.insize,
False)
# These iterators load the images with subprocesses running in parallel
# to the training/validation.
train_iter = chainer.iterators.MultiprocessIterator(
train, args.batchsize, n_processes=args.loaderjob)
val_iter = chainer.iterators.MultiprocessIterator(
val, args.val_batchsize, repeat=False, n_processes=args.loaderjob)
converter = dataset.concat_examples
# Set up an optimizer
optimizer = chainer.optimizers.MomentumSGD(lr=0.01, momentum=0.9)
if args.nvtx_mark:
optimizer = create_marked_profile_optimizer(optimizer,
sync=True,
sync_level=2)
optimizer.setup(model)
# Set up a trainer
updater = training.updaters.StandardUpdater(train_iter,
optimizer,
converter=converter,
device=args.gpu)
if args.iter > 0:
# Evaluation and etc are skipped during profiling
# because these functionalities are not necessary.
trainer = training.Trainer(updater, (args.iter, 'iteration'),
out=args.out)
else:
trainer = training.Trainer(updater, (args.epoch, 'epoch'), args.out)
val_interval = (1 if args.test else 100000), 'iteration'
log_interval = (1 if args.test else 1000), 'iteration'
trainer.extend(extensions.Evaluator(val_iter,
model,
converter=converter,
device=args.gpu),
trigger=val_interval)
trainer.extend(extensions.dump_graph('main/loss'))
trainer.extend(extensions.snapshot(), trigger=val_interval)
trainer.extend(extensions.snapshot_object(
model, 'model_iter_{.updater.iteration}'),
trigger=val_interval)
# Be careful to pass the interval directly to LogReport
# (it determines when to emit log rather than when to read observations)
trainer.extend(extensions.LogReport(trigger=log_interval))
trainer.extend(extensions.observe_lr(), trigger=log_interval)
trainer.extend(extensions.PrintReport([
'epoch', 'iteration', 'main/loss', 'validation/main/loss',
'main/accuracy', 'validation/main/accuracy', 'lr'
]),
trigger=log_interval)
trainer.extend(extensions.ProgressBar(update_interval=10))
if args.resume:
chainer.serializers.load_npz(args.resume, trainer)
trainer.run()