def main(epochs,
enable_function,
buffer_size,
batch_size,
mode,
growth_rate,
output_classes,
depth_of_model=None,
num_of_blocks=None,
num_layers_in_each_block=None,
data_format='channels_last',
bottleneck=True,
compression=0.5,
weight_decay=1e-4,
dropout_rate=0.,
pool_initial=False,
include_top=True,
train_mode='custom_loop',
data_dir=None):
train_obj = Train(epochs, enable_function)
train_dataset, test_dataset = create_dataset(buffer_size, batch_size,
data_format, data_dir)
model = densenet.DenseNet(mode, growth_rate, output_classes,
depth_of_model, num_of_blocks,
num_layers_in_each_block, data_format,
bottleneck, compression, weight_decay,
dropout_rate, pool_initial, include_top)
print('Training...')
if train_mode == 'custom_loop':
return train_obj.custom_loop(train_dataset, test_dataset, model)
elif train_mode == 'keras_fit':
return train_obj.keras_fit(train_dataset, test_dataset, model)
def test_one_epoch_with_keras_fit(self):
epochs = 1
enable_function = True
depth_of_model = 7
growth_rate = 2
num_of_blocks = 3
output_classes = 10
mode = 'from_depth'
data_format = 'channels_last'
train_dataset = create_sample_dataset(batch_size=1)
test_dataset = create_sample_dataset(batch_size=1)
train_obj = train.Train(epochs, enable_function)
model = densenet.DenseNet(mode, growth_rate, output_classes,
depth_of_model, num_of_blocks, data_format)
train_obj.keras_fit(train_dataset, test_dataset, model)