def test(self, test_data: List[GraphData], best_model_file: Optional[str] = None) -> Tuple[Loss, MicroF1]:
print("##### Test Model #####")
with torch.no_grad():
if best_model_file:
self._net.load_state_dict(torch.load(best_model_file))
self._net.eval()
results = []
for graph in test_data:
adjs = [sample_neighbors(graph.adj_coo_matrix,size) for size in [25,10]]
output = self._net(graph.features_vectors,
*adjs)
labels = graph.labels
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
result = Loss("Test Loss", avg_loss), MicroF1("Test F1", avg_f1)
print(f"{result[0]}, {result[1]}")
return result
def test(self,
test_data: List[GraphData],
best_model_file: Optional[str] = None) -> Tuple[Loss, MicroF1]:
print("##### Test Model #####")
with torch.no_grad():
if best_model_file:
self._net.load_state_dict(torch.load(best_model_file))
self._net.eval()
train_embeddings, train_labels = self._calc_embeddings(
self._train_data)
test_embeddings, test_labels = self._calc_embeddings(test_data)
log = MultiOutputClassifier(SGDClassifier(loss="log"), n_jobs=10)
log.fit(train_embeddings.detach().cpu(),
train_labels.detach().cpu())
results = []
for graph in test_data:
output = self._run_net(graph)
loss = self.unsup_loss(output, graph.positive_pairs,
graph.adj_coo_matrix)
results.append((loss.item(), ))
avg_loss = list(self._avg_results(results))[0]
f1 = MicroF1.calc(
torch.from_numpy(log.predict(test_embeddings.detach().cpu())),
test_labels.detach().cpu())
result = Loss("Test Loss", avg_loss), MicroF1("Test F1", f1)
print(f"{result[0]}, {result[1]}")
return result
def test(self, test_data: GraphData, best_model_file: Optional[str] = None) -> Tuple[Loss, MicroF1]:
print("##### Test Model #####")
loader = MiniBatchLoader(
test_data.adj_coo_matrix, [25,10], test_data.test_mask, batch_size=512, shuffle=False)
with torch.no_grad():
if best_model_file:
self._net.load_state_dict(torch.load(best_model_file))
self._net.eval()
results = []
for batch_step in tqdm(loader):
batch_step = batch_step.to(self._current_device)
output = self._net(
test_data.features_vectors[batch_step.batch_idxs].to(self._current_device), *batch_step.sampled_adjs)
output = output[batch_step.sampled_idxs]
labels = test_data.labels[batch_step.target_idxs].to(self._current_device)
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
result = Loss("Test Loss", avg_loss), MicroF1("Test F1", avg_f1)
print(f"{result[0]}, {result[1]}")
return result
def _evaluate(self) -> Tuple[Loss, MicroF1]:
with torch.no_grad():
self._net.eval()
results = []
for batch_step in tqdm(self._validation_loader):
batch_step = batch_step.to(self._current_device)
output = self._net(
self._validation_data.features_vectors[
batch_step.batch_idxs].to(self._current_device),
*batch_step.sampled_adjs)
output = output[batch_step.sampled_idxs]
labels = self._validation_data.labels[
batch_step.target_idxs].to(self._current_device)
loss = self.unsup_loss(output,
self._validation_data.positive_pairs,
self._validation_data.adj_coo_matrix)
f1 = MicroF1.calc(output, labels)
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Validation Loss",
avg_loss), MicroF1("Validation F1", avg_f1)
def _evaluate(self) -> Tuple[Loss, MicroF1]:
with torch.no_grad():
self._net.eval()
results = []
for step in self._validation_data:
output = self._net(step.features_vectors, step.adj_coo_matrix)
results.append((
self._loss_fn(output, step.labels).item(),
#accuracy(output, step.labels),
MicroF1.calc(output, step.labels),
))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Validation Loss",
avg_loss), MicroF1("Validation F1", avg_f1)
def _evaluate(self) -> Tuple[Loss, MicroF1]:
with torch.no_grad():
self._net.eval()
results = []
for graph in self._validation_data:
adjs = [sample_neighbors(graph.adj_coo_matrix,size) for size in [25,10]]
output = self._net(graph.features_vectors,
*adjs)
labels = graph.labels
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Validation Loss", avg_loss), MicroF1("Validation F1", avg_f1)
def _evaluate(self) -> Tuple[Loss, MicroF1]:
with torch.no_grad():
self._net.eval()
results = []
for graph in self._validation_data:
output = self._run_net(graph)
labels = graph.labels
loss = self.unsup_loss(output, graph.positive_pairs,
graph.adj_coo_matrix)
f1 = 0
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Validation Loss",
avg_loss), MicroF1("Validation F1", avg_f1)
def _train(self) -> Tuple[Loss, MicroF1]:
self._net.train()
results = []
for train_graph in self._train_data:
self._optim.zero_grad()
output = self._run_net(train_graph)
labels = train_graph.labels
loss = self.unsup_loss(output, train_graph.positive_pairs,
train_graph.adj_coo_matrix)
f1 = 0
loss.backward()
torch.nn.utils.clip_grad_value_(self._net.parameters(), 5)
self._optim.step()
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Train Loss", avg_loss), MicroF1("Train F1", avg_f1)
def _train(self) -> Tuple[Loss, MicroF1]:
self._net.train()
results = []
for train_graph in tqdm(self._train_data):
self._optim.zero_grad()
adjs = [sample_neighbors(train_graph.adj_coo_matrix,size) for size in [25,10]]
output = self._net(train_graph.features_vectors,
*adjs)
labels = train_graph.labels
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
loss.backward()
torch.nn.utils.clip_grad_value_(self._net.parameters(), 5)
self._optim.step()
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Train Loss", avg_loss), MicroF1("Train F1", avg_f1)
def _train(self) -> Tuple[Loss, MicroF1]:
self._net.train()
results = []
for batch_step in tqdm(self._train_loader):
batch_step = batch_step.to(self._current_device)
self._optim.zero_grad()
output = self._net(
self._train_data.features_vectors[batch_step.batch_idxs].to(self._current_device), *batch_step.sampled_adjs)
output = output[batch_step.sampled_idxs]
labels = self._train_data.labels[batch_step.target_idxs].to(self._current_device)
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
loss.backward()
self._optim.step()
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Train Loss", avg_loss), MicroF1("Train F1", avg_f1)
def test(self,
test_data: List[GraphData],
best_model_file: Optional[str] = None) -> Tuple[Loss, MicroF1]:
print("##### Test Model #####")
with torch.no_grad():
if best_model_file:
self._net.load_state_dict(torch.load(best_model_file))
self._net.eval()
results = []
for step in test_data:
output = self._net(step.features_vectors, step.adj_coo_matrix)
results.append((
self._loss_fn(output, step.labels).item(),
MicroF1.calc(output, step.labels),
))
avg_loss, avg_F1 = self._avg_results(results)
result = Loss("Test Loss", avg_loss), MicroF1("Test F1", avg_F1)
print(f"{result[0]}, {result[1]}")
return result
def _train(self) -> Tuple[Loss, MicroF1]:
self._net.train()
results = []
for batch in self._batch_train_data:
self._optim.zero_grad()
labels = torch.cat([data.labels for data in batch])
output = torch.cat([
self._net(data.features_vectors, data.adj_coo_matrix)
for data in batch
])
loss = self._loss_fn(output, labels)
f1 = MicroF1.calc(output, labels)
loss.backward()
self._optim.step()
results.append((loss.item(), f1))
avg_loss, avg_f1 = self._avg_results(results)
return Loss("Train Loss", avg_loss), MicroF1("Train F1", avg_f1)