def run_(data,
dataset,
edge_index,
train_node,
writer,
runs=10,
lr=0.01,
weight_decay=0.005):
model = Net(dataset)
losses, accs = [], []
self_loop = torch.stack((torch.LongTensor(range(
data.num_nodes)), torch.LongTensor(range(data.num_nodes))),
dim=0).cuda()
# pdb.set_trace()
model.to(device).reset_parameters()
optimizer = Adam(model.parameters(), lr=0.01, weight_decay=0.0005)
criterion = nn.BCEWithLogitsLoss(torch.ones(data.num_class).cuda())
for i in range(1000):
# train(model, optimizer, data, self_loop)
train(model, optimizer, data, edge_index, criterion, writer, i)
if torch.cuda.is_available():
torch.cuda.synchronize()
# loss, acc = evaluate(model, data, self_loop)
loss, acc = evaluate(model, data, edge_index, criterion)
print('w/o Val Loss: {:.4f}, Test Accuracy: {:.3f}'.format(loss, acc))
return loss, acc
def run_(data,
dataset,
edge_index,
train_node,
runs=10,
lr=0.01,
weight_decay=0.005):
model = Net(dataset)
losses, accs = [], []
model.to(device).reset_parameters()
optimizer = Adam(model.parameters(), lr=0.01, weight_decay=0.0005)
for i in range(200):
train(model, optimizer, data, edge_index)
if torch.cuda.is_available():
torch.cuda.synchronize()
loss, acc = evaluate(model, data, edge_index)
losses.append(loss)
accs.append(acc)
print('Val Loss: {:.4f}, Test Accuracy: {:.3f}'.format(loss, acc))
losses, accs = tensor(losses), tensor(accs)
print(
'::::::No Edge::::::\n Val Loss: {:.4f}, Mean Test Accuracy: {:.3f} ± {:.3f}'
.format(losses.mean().item(),
accs.mean().item(),
accs.std().item()))