help='Number of hidden units.')
parser.add_argument('--dropout',
type=float,
default=0.5,
help='Dropout rate (1 - keep probability).')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data()
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=args.weight_decay)
# optimizer = optim.Adam(model.parameters(),
# lr=args.lr, weight_decay=10)
if args.cuda:
model.cuda()
features = features.cuda()
parser.add_argument('--hidden', type=int, default=16,
help='Number of hidden units.')
parser.add_argument('--dropout', type=float, default=0.5,
help='Dropout rate (1 - keep probability).')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
print(f"参数列表:{args}")
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(
path='./data/cora/')
print(adj.shape, features.shape, labels.shape)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
if args.cuda:
model.cuda()
features = features.cuda()
adj = adj.cuda()
labels = labels.cuda()
action='store_true',
default=False,
help='Using Batch Normalization')
#dataset = 'citeseer'
#dataset = 'pubmed'
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
torch.cuda.set_device(args.cuda_device)
dataset = args.dataset
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
A, features, labels, idx_train, idx_val, idx_test, adj_mask = load_data(
dataset)
idx_unlabel = torch.range(idx_train.shape[0], labels.shape[0] - 1, dtype=int)
features = features.cuda()
adj_mask = adj_mask.cuda()
print(features[adj_mask[0]].shape)
assert False
# Model and optimizer
model = MLP(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
input_droprate=args.input_droprate,
hidden_droprate=args.hidden_droprate,
use_bn=args.use_bn)
optimizer = optim.Adam(model.parameters(),
parser.add_argument('--neg_sample_weight', type=int, default=20,
help='Negative sample size')
parser.add_argument('--transfer', action='store_true', default=False,
help='Transfer learning - using smaller learning rate when transfering')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, adj_ds, neg_adj_list, train_edges, degrees, features, labels, idx_train, idx_val, idx_test = load_data(args.path, args.dataset)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
optimizer2 = optim.Adam(
[
{"params": model.gc1.parameters(), "lr": args.lr/10},
{"params": model.gc2.parameters(), "lr": args.lr}
],
lr=args.lr,
weight_decay=args.weight_decay
class args_:
def __init__(self):
self.hidden=16
self.no_cuda=False
self.fastmode=False
self.seed=42
self.epochs=200
self.lr=0.01
self.weight_decay=5e-4
self.dropout=0.5
# this does not matter at all.
# whatever.
args=args_()
args.cuda = not args.no_cuda and torch.cuda.is_available()
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(path="/root/AGI/lazero/brainfuck/pygcn/data/cora/")
# you'd better see this.
# idx is for index.
# active: 10:00 AM -> 12:00 PM
# 12:00 noon <-> 2:00 AM
# mod operation.
# how to let computer calc this?
# you can assign random things.
# Model and optimizer
# anyway, do you want to train some letters? the network made up of letters.
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
help='Weight decay (L2 loss on parameters).')
parser.add_argument('--hidden', type=int, default=16,
help='Number of hidden units.')
parser.add_argument('--dropout', type=float, default=0.5,
help='Dropout rate (1 - keep probability).')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(path="data/cora/", dataset="cora")
def test():
model.eval()
output = model(features, adj)
loss_test = F.nll_loss(output[idx_test], labels[idx_test])
acc_test = accuracy(output[idx_test], labels[idx_test])
print("Test set results:",
"loss= {:.4f}".format(loss_test.item()),
"accuracy= {:.4f}".format(acc_test.item()))
def train(epoch):
t = time.time()
model.train()
optimizer.zero_grad()
output = model(features, adj)
self.hidden = 16
self.no_cuda = False
self.fastmode = False
self.seed = 42
self.epochs = 200
self.lr = 0.01
self.weight_decay = 5e-4
self.dropout = 0.5
# this does not matter at all.
# whatever.
args = args_()
args.cuda = not args.no_cuda and torch.cuda.is_available()
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(
path="pygcn/data/cora/")
# first is adj matrix.
# second??
print(adj.shape, features.shape, labels.shape, idx_train.shape, idx_val.shape,
idx_test.shape)
# torch.Size([2708, 2708]) torch.Size([2708, 1433]) torch.Size([2708]) torch.Size([140]) torch.Size([300]) torch.Size([1000])
print(type(adj), type(features), type(labels), type(idx_train), type(idx_val),
type(idx_test))
# pdd={"adj":adj, "features":features, "labels":labels, "idx_train":idx_train, "idx_val":idx_val, "idx_test":idx_test}
# for x in pdd.keys():
# print(labels)
# # print(x.__name__)
# print(x,pdd[x].shape)
# print(pdd[x])
# you'd better see this.
# # Model and optimizer
#os.environ["CUDA_VISIBLE_DEVICES"] = "5"
torch.cuda.set_device(5)
print('args.cuda', args.cuda)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
print(args.seed)
# Load data
dataset = 'cora'
#dataset = 'citeseer'
# dataset = 'pubmed'
adj, A, _, features, labels, idx_train, idx_val, idx_test, edges, adj_ = load_data(
dataset)
model = NGCN(nfeat_1=features.shape[1],
nfeat_2=features.shape[1],
nhid_1=args.hid1,
nhid_2=args.hid1,
nclass=labels.max().item() + 1,
dropout=args.dropout)
# dropout=0.5)
# center_loss_1 = CenterLoss(num_classes=labels.max().item() + 1, feat_dim=16, use_gpu=args.cuda)
# center_loss_2 = CenterLoss(num_classes=labels.max().item() + 1, feat_dim=16, use_gpu=args.cuda)
params = list(model.parameters(
)) #+ list(center_loss_1.parameters()) + list(center_loss_2.parameters())
args = parser.parse_args()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
# Load data
dataset = []
graphs.save_numpy(
r'C:\Users\Pasi\OneDrive\Documents\Uni\MSem. 4 - SS 20\MT - Master Thesis\Simulator and Models\MT_SimpleDataGenerator\pygcn\data\sdg_fraud\\',
['price', 'new_value', 'old_value'])
for graph in graphs.get_raw_list():
adj, features, labels, idx_train, idx_val, idx_test = load_data(
path="../pygcn/data/sdg_fraud/",
dataset=graph.get_name(),
train_size=len(graph) - 1,
validation_size=0)
dataset.append(
Bunch(
name=graph.get_name(),
adj=adj,
features=features,
# labels_raw=labels,
labels=readout_labels(labels, args.classification_mode),
idx_train=idx_train,
idx_val=idx_val,
idx_test=idx_test))
random.shuffle(dataset)
help='Number of hidden units.')
parser.add_argument('--dropout',
type=float,
default=0.5,
help='Dropout rate (1 - keep probability).')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(
path="./data/cora/", dataset='cora', train_size=400, validation_size=50)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=args.weight_decay)
if args.cuda:
model.cuda()
features = features.cuda()
adj = adj.cuda()
labels = labels.cuda()
parser.add_argument('--dropout',
type=float,
default=0.5,
help='Dropout rate (1 - keep probability).')
parser.add_argument('--dataset', type=str, default='cora')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(
dataset=args.dataset)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=args.weight_decay)
if args.cuda:
model.cuda()
features = features.cuda()
adj = adj.cuda()
labels = labels.cuda()
parser.add_argument('--dropout', type=float, default=0.5,
help='Dropout rate (1 - keep probability).')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
np.random.seed(187348292)
# Load data
# adj, features, labels, idx_train, idx_val, idx_test = load_data()
adj, features, labels = load_data()
adj_csr = sp.csr_matrix(adj)
idxs = np.arange(adj.shape[0])
train_size = 500
val_size = 300
test_size = 1000
idx_train, idx_val, idx_test = training_split(idxs, train_size, val_size, test_size)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
print("start")
# Load data
graph = np.load("D:/data/graph/graph_save.npy")
# Load data
node = torch.load("D:/data/node/node.pt")
with open("D:/data/node/nodelist.json", 'r') as fp:
node_list = json.load(fp)
print("Data Load")
adj, features, labels, idx_train, idx_val, idx_test = load_data(graph[0], node, node_list)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
if args.cuda:
model.cuda()
features = features.cuda()
adj = adj.cuda()
labels = labels.cuda()
idx_train = idx_train.cuda()
help='Hard assignment of gumbel softmax')
parser.add_argument('--beta',
type=float,
default=0,
help='Beta param of gumbel softmax, default=0')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, adj_ds, _, _, _, features, labels, idx_train, idx_val, idx_test = load_data(
args.path, args.dataset)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=args.weight_decay)
optimizer2 = optim.Adam([{
"params": model.gc1.parameters(),
"lr": args.lr / 10
}, {
"params": model.gc2.parameters(),
"lr": args.lr
help='Dropout rate (1 - keep probability).')
parser.add_argument('--dataset', type=str, default="cora",
help='Dataset (cora...)')
parser.add_argument('--noramlize_features', type=bool, default=True,
help='normalize features')
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data("../data/%s/" % args.dataset, args.dataset, args.noramlize_features)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout)
optimizer = optim.Adam(model.parameters(),
lr=args.lr, weight_decay=args.weight_decay)
if args.cuda:
model.cuda()
features = features.cuda()
adj = adj.cuda()
labels = labels.cuda()
idx_train = idx_train.cuda()
help='Dropout rate (1 - keep probability).')
parser.add_argument('--data_type',
type=str,
default="cora",
help="Type of dataset to train and eval.")
args = parser.parse_args()
args.cuda = not args.no_cuda and torch.cuda.is_available()
np.random.seed(args.seed)
torch.manual_seed(args.seed)
if args.cuda:
torch.cuda.manual_seed(args.seed)
# Load data
adj, features, labels, idx_train, idx_val, idx_test = load_data(args.data_type)
# Model and optimizer
model = GCN(nfeat=features.shape[1],
nhid=args.hidden,
nclass=labels.max().item() + 1,
dropout=args.dropout,
data_type=args.data_type)
if args.data_type == "cora":
loss = LabelSmoothLoss()
elif args.data_type.startswith("elliptic"):
loss = nn.CrossEntropyLoss(weight=torch.Tensor([0.7, 0.3, 0.0]))
optimizer = optim.Adam(model.parameters(),
lr=args.lr,
weight_decay=args.weight_decay)