decay=0.00001,
amsgrad=True)
# In[12]:
csv_logger = CSVLogger(csvlogger_name, append=True, separator=',')
checkpoint = ModelCheckpoint(checkpoint_filename,
monitor='val_loss',
verbose=1,
save_best_only=False,
period=1)
earlystopping = EarlyStopping(patience=5, mode='min')
#callbacks = [tensorboard, checkpoint]
callbacks = [checkpoint, csv_logger, earlystopping]
# In[13]:
model.compile(loss='mae', optimizer=optimizer, metrics=['mse'])
# In[14]:
model.fit_generator(train_generator,
epochs=epochs,
steps_per_epoch=steps_per_epoch,
validation_data=validation_generator,
validation_steps=validation_steps,
callbacks=callbacks)
root = './data'
download = True # download MNIST dataset or not
trans = transforms.Compose(
[transforms.ToTensor(), transforms.Normalize((0.5,), (1.0,))])
train_set = dset.MNIST(root=root, train=True,
transform=trans, download=download)
test_set = dset.MNIST(root=root, train=False, transform=trans)
model = LeNet()
criterion = nn.CrossEntropyLoss()
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.9)
model.compile(optimizer, criterion, metrics=['top1', 'top2'])
elif test_type == 'multi tasks classification':
train_set = CategoricalDatasetMultiTasks(N, D_in, D_out1, D_out2)
test_set = CategoricalDatasetMultiTasks(
int(N * 0.25), D_in, D_out1, D_out2)
model = MultiTasksClassification(D_in, H1, H2, D_out1, D_out2)
optimizer = optim.SGD(model.parameters(), lr=0.01, momentum=0.9)
model.compile(optimizer, multiCrossEntropy, metrics=[
'top1', 'top2'], multi_tasks=['output_a', 'output_b'])
train_loader = torch.utils.data.DataLoader(
