def evaluate(self, samples=1000):
"""
Evaluate after training procedure finished
:return: A log that contains information about validation
"""
Outputs = torch.zeros(self.test_data_loader.n_samples,
self.model.num_classes, samples).to(self.device)
targets = torch.zeros(self.test_data_loader.n_samples)
self.model.eval()
with torch.no_grad(
): # torch.no_grad() 是一个上下文管理器，被该语句 wrap 起来的部分将不会track 梯度。
start = 0
for batch_idx, (data, target) in enumerate(self.test_data_loader):
end = len(data) + start
data, target = data.to(self.device), target.to(self.device)
loss = 0
outputs = torch.zeros(data.shape[0], self.model.num_classes,
samples).to(self.device)
if samples == 1:
out, _ = self.model(data)
loss = self.criterion(out, target)
outputs[:, :, 0] = out
elif samples > 1:
mlpdw_cum = 0
for i in range(samples):
out, _ = self.model(data, sample=True)
mlpdw_i = self.criterion(out, target)
mlpdw_cum = mlpdw_cum + mlpdw_i
outputs[:, :, i] = out
mlpdw = mlpdw_cum / samples
loss = mlpdw
Outputs[start:end, :, :] = outputs
targets[start:end] = target
start = end
self.test_metrics.update('loss', loss.item())
for met in self.metric_ftns:
self.test_metrics.update(met.__name__,
met(outputs, target, type="VD"))
result = self.test_metrics.result()
# print logged informations to the screen
for key, value in result.items():
self.logger.info(' {:15s}: {}'.format(str(key), value))
# self._visualization(Outputs, targets)
test_uncertainities(Outputs,
targets,
self.model.num_classes,
self.logger,
save_path=str(self.result_dir))
def evaluate(self):
"""
Evaluate after training procedure finished
:return: A log that contains information about validation
"""
Outputs = torch.zeros(self.test_data_loader.n_samples,
self.models[0].num_classes, self.n_ensembles)
global targets
for i, model in enumerate(self.models):
model.eval()
outputs = torch.zeros(self.test_data_loader.n_samples,
model.num_classes).to(self.device)
targets = torch.zeros(self.test_data_loader.n_samples)
with torch.no_grad(
): # torch.no_grad() 是一个上下文管理器，被该语句 wrap 起来的部分将不会track 梯度。
start = 0
for batch_idx, (data,
target) in enumerate(self.test_data_loader):
end = len(data) + start
data, target = data.to(self.device), target.to(self.device)
output = model(data)
outputs[start:end, :] = output
targets[start:end] = target
start = end
loss = self.criterion(output, target)
self.test_metrics.update('loss_' + str(i), loss.item())
for met in self.metric_ftns:
self.test_metrics.update(
met.__name__ + '_' + str(i),
met(output, target, type="DE"))
self._visualization(outputs, targets, i)
Outputs[:, :, i] = outputs
self._info_ensemble(Outputs, targets)
result = self.test_metrics.result()
# print logged informations to the screen
for key, value in result.items():
self.logger.info(' {:15s}: {}'.format(str(key), value))
# self._visualization_ensemble(Outputs, targets)
test_uncertainities(Outputs,
targets,
self.models[0].num_classes,
self.logger,
save_path=str(self.result_dir))