def sgd_optimize_logreg(datasets):
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of [int] labels
classifier = LogisticRegression(input=x, n_in=28 * 28, n_out=10)
trainer = SGDTrainer(classifier, datasets)
trainer.build(x, y)
return trainer.train()
def sgd_optimize_mlp(datasets):
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of [int] labels
classifier = MLP(input=x, n_in=28 * 28, n_out=10, n_hiddens=[1500])
trainer = SGDTrainer(classifier, datasets, learning_rate=0.01, L1_reg=0.0001,
L2_reg=0.001, n_epochs=500, batch_size=20)
trainer.build(x, y)
return trainer.train()
def sgd_optimize_lenet(datasets):
x = T.matrix('x') # the data is presented as rasterized images
y = T.ivector('y') # the labels are presented as 1D vector of [int] labels
classifier = LeNet(input=x, nkerns=[20, 50], filter_shapes=[[5, 5], [5, 5]],
image_shapes=[[28, 28], [12, 12]], batch_size=500,
n_hidden=[500], n_out=10)
trainer = SGDTrainer(classifier, datasets, learning_rate=0.01,
L1_reg=0.0001, L2_reg=0.001, n_epochs=100,
batch_size=classifier.batch_size)
trainer.build(x, y)
return trainer.train()
def main():
parser = argparse.ArgumentParser(description="ReID Baseline Training")
parser.add_argument("--config_file",
default="",
help="path to config file",
type=str)
parser.add_argument("opts",
help="Modify config options using the command-line",
default=None,
nargs=argparse.REMAINDER)
args = parser.parse_args()
if args.config_file != "":
cfg.merge_from_file(args.config_file)
cfg.merge_from_list(args.opts)
cfg.freeze()
output_dir = cfg.OUTPUT_DIR
if output_dir and not os.path.exists(output_dir):
os.makedirs(output_dir)
num_gpus = torch.cuda.device_count()
logger = setup_logger('reid_baseline', output_dir, 0)
logger.info('Using {} GPUS'.format(num_gpus))
logger.info('Running with config:\n{}'.format(cfg))
train_dl, val_dl, num_query, num_classes = make_dataloader(cfg, num_gpus)
model = build_model(cfg, num_classes)
loss = make_loss(cfg, num_classes)
trainer = SGDTrainer(cfg, model, train_dl, val_dl, loss, num_query,
num_gpus)
logger.info('train transform: \n{}'.format(train_dl.dataset.transform))
logger.info('valid transform: \n{}'.format(val_dl.dataset.transform))
logger.info(type(model))
logger.info(loss)
logger.info(trainer)
for epoch in range(trainer.epochs):
for batch in trainer.train_dl:
trainer.step(batch)
trainer.handle_new_batch()
trainer.handle_new_epoch()
