clean code

LRC/learning_rate.py

@@ -76,4 +76,3 @@ def cosine_with_warmup_decay(learning_rate, lr_min, steps_one_epoch,
             fluid.layers.assign(cosine_lr, lr)
 
     return lr
-

LRC/model.py

@@ -175,7 +175,6 @@ def StemConv(input, C_out, kernel_size, padding):
     return bn_a
 
 
-
 class NetworkCIFAR(object):
     def __init__(self, C, class_num, layers, auxiliary, genotype):
         self._layers = layers

LRC/reader_cifar.py

@@ -52,6 +52,7 @@ half_length = 8
 CIFAR_MEAN = [0.49139968, 0.48215827, 0.44653124]
 CIFAR_STD = [0.24703233, 0.24348505, 0.26158768]
 
+
 def generate_reshape_label(label, batch_size, CIFAR_CLASSES=10):
     reshape_label = np.zeros((batch_size, 1), dtype='int32')
     reshape_non_label = np.zeros(
@@ -153,10 +154,6 @@ def reader_creator_filepath(filename, sub_name, is_training, args):
             if len(batch_data) == args.batch_size:
                 batch_data = np.array(batch_data, dtype='float32')
                 batch_label = np.array(batch_label, dtype='int64')
-#
-#                batch_data = pickle.load(open('input.pkl'))
-#                batch_label = pickle.load(open('target.pkl')).reshape(-1,1)
-#               
                 if is_training:
                     flatten_label, flatten_non_label = \
                       generate_reshape_label(batch_label, args.batch_size)

LRC/train_mixup.py

@@ -70,6 +70,7 @@ dataset_train_size = 50000.
 image_size = 32
 genotypes.DARTS = genotypes.MY_DARTS_list[args.model_id]
 
+
 def main():
     image_shape = [3, image_size, image_size]
     devices = os.getenv("CUDA_VISIBLE_DEVICES") or ""
@@ -79,7 +80,8 @@ def main():
     model = Network(args.init_channels, CIFAR_CLASSES, args.layers,
                     args.auxiliary, genotype)
     
-    steps_one_epoch = math.ceil(dataset_train_size / (devices_num * args.batch_size))
+    steps_one_epoch = math.ceil(dataset_train_size / 
+                                (devices_num * args.batch_size))
     train(model, args, image_shape, steps_one_epoch)
 
 
@@ -136,13 +138,6 @@ def train(model, args, im_shape, steps_one_epoch):
             main_program=train_prog,
             predicate=if_exist)
 
-    #if args.pretrained_model:
-
-    #    def if_exist(var):
-    #        return os.path.exists(os.path.join(args.pretrained_model, var.name))
-
-    #    fluid.io.load_vars(exe, args.pretrained_model, main_program=train_prog, predicate=if_exist)
-
     exec_strategy = fluid.ExecutionStrategy()
     exec_strategy.num_threads = 1
     build_strategy = fluid.BuildStrategy()

LRC/utils.py

@@ -34,10 +34,6 @@ def mixup_data(x, y, batch_size, alpha=1.0):
         lam = 1.
     index = np.random.permutation(batch_size)
 
-    #
-    #lam = 0.5
-    #index = np.arange(batch_size-1, -1, -1)
-    #
     mixed_x = lam * x + (1 - lam) * x[index, :]
     y_a, y_b = y, y[index]
     return mixed_x.astype('float32'), y_a.astype('int64'),\