def _post_clone_action(self, **kwargs):
from datetime import datetime
model = self._get_details_query(None, **kwargs)
name = "{0} clone: {1}".format(
model.name,
datetime.now().strftime("%Y-%m-%d %H:%M:%S"))
new_model = Model(name=name,
train_import_handler=model.train_import_handler,
test_import_handler=model.test_import_handler)
new_model.save()
new_model.classifier = model.classifier
new_model.features_set.from_dict(model.features_set.features,
commit=False)
new_model.tags = model.tags
new_model.save()
for tag in new_model.tags:
tag.update_counter()
return self._render(
{
self.OBJECT_NAME: new_model,
'status': 'New model "{0}" created'.format(new_model.name)
},
code=201)
idx = np.arange(600)
np.random.RandomState(seed=124).shuffle(idx)
idx_train, idx_test = idx[:480], idx[480:]
model = Model(pool_size=int(max_num_nodes * 0.25), device=device).to(device)
model.train()
# optimizer = optim.SGD(model.parameters(), lr=1e-5)
optimizer = optim.Adam(model.parameters())
for e in tqdm(range(args.epochs)):
pred_labels = []
for i, idx in enumerate(idx_train):
adj_train = torch.from_numpy(adj[idx]).to(device).float()
features_train = torch.from_numpy(features[idx]).to(device).float()
labels_train = labels[idx].view(-1).long()
graph_feat = model(features_train, adj_train)
output = model.classifier(graph_feat)
criterion = nn.CrossEntropyLoss()
if args.link_pred:
loss = criterion(output, labels_train) + model.link_pred_loss()
else:
loss = criterion(output, labels_train)
loss.backward()
pred_labels.append(output.argmax())
if i % 32 == 0:
optimizer.step()
optimizer.zero_grad()
optimizer.step()
optimizer.zero_grad()
pred_labels = torch.stack(pred_labels, dim=0)
acc = (pred_labels.long() == labels[idx_train].long()).float().mean()
