x_train, y_train, x_test, y_test = data
m.set_data((x_train, y_train), (x_test, y_test))
m.train(ROUND)
def train_gbdt_single(data, meta):
depth = cfg.Depth[args.mode][args.data]
lr = cfg.Learning_rate[args.mode][args.data]
p = {'max_depth': depth, 'max_leaves': int(0.75 * 2 ** depth), 'topk': 0, 'loss': b"ce_column",
'gamma': GAMMA, 'num_threads': num_threads, 'max_bins': meta['bin'], 'lr': lr, 'reg_l2': 1.0,
'early_stop': 25, 'one_side': True, 'verbose': True,
'min_samples': min_samples}
x_train, y_train, x_test, y_test = data
m = GBDTSingle(LIB, out_dim=meta['out'], params=p)
m.set_data((x_train, y_train), (x_test, y_test))
m.train_multi(ROUND)
if __name__ == '__main__':
m = DataLoader()
data, meta = m.get(args.data, False, args.N)
if args.mode == 'gbdtmo':
train_gbdt_multi(data, meta)
elif args.mode == 'gbdtso':
train_gbdt_single(data, meta)
elif args.mode == 'lightgbm':
train_lightgbm(data, meta)
x_train, y_train, x_test, y_test = data
m.set_data((x_train, y_train), (x_test, y_test))
t = time.time()
m.train(ROUND)
t = time.time() - t
if args.time == 1:
print("Average time: {:.3f}".format(t / ROUND))
else:
print("Total time: {:.3f}".format(t))
del m
if __name__ == '__main__':
m = DataLoader()
if args.data == 'mnist_reg':
data, meta = m.get('mnist_reg', True, 10000)
for k in [4, 8, 16]:
regression(data, meta, 7, 0.1, k, True)
regression(data, meta, 7, 0.1, k, False)
elif args.data == 'nus-wide':
data, meta = m.get('nus-wide', True, 10000)
for k in [8, 16, 32, 64]:
regression(data, meta, 8, 0.1, k, True)
regression(data, meta, 8, 0.1, k, False)
elif args.data == 'Caltech101':
data, meta = m.get('Caltech101', False, 10000)
for k in [8, 16, 32, 64]:
classification(data, meta, 8, 0.1, k, True)
classification(data, meta, 8, 0.1, k, False)
elif args.data == 'mnist':
data, meta = m.get('mnist', False, 10000)
if __name__ == '__main__':
p = {
'mnist': [8, 0.1],
'Caltech101': [9, 0.1],
'mnist_reg': [7, 0.1],
'nus-wide': [8, 0.1],
}
name = args.data
m = DataLoader()
out_multi, out_single = 0.0, 0.0
for i in range(3):
if name == 'mnist':
data, meta = m.get(name, False, 10000)
out_multi += classification_multi(data, meta, p[name][0],
p[name][1])
out_single += classification_single(data, meta, p[name][0],
p[name][1])
elif name == 'Caltech101':
data, meta = m.get(name, False, 10000)
out_multi += classification_multi(data, meta, p[name][0],
p[name][1])
out_single += classification_single(data, meta, p[name][0],
p[name][1])
elif name == 'mnist_reg':
data, meta = m.get(name, True, 10000)
out_multi += regression_multi(data, meta, p[name][0], p[name][1])
out_single += regression_single(data, meta, p[name][0], p[name][1])
elif name == 'nus-wide':
