Delete some redundant

Author: Laicheng0830

examples/basic_tutorials/tutorial_using_tensorboradX.py

@@ -40,11 +40,27 @@
 from tensorlayerx.nn import Module
 from tensorlayerx.nn import Linear, Dropout, BatchNorm1d
 from tensorboardX import SummaryWriter
-
+from tensorlayerx.dataflow import Dataset, DataLoader
 
 X_train, y_train, X_val, y_val, X_test, y_test = tlx.files.load_mnist_dataset(shape=(-1, 784))
 
 
+class mnistdataset(Dataset):
+
+    def __init__(self, data=X_train, label=y_train):
+        super().__init__()
+        self.data = data
+        self.label = label
+
+    def __getitem__(self, index):
+        data = self.data[index].astype('float32')
+        label = self.label[index].astype('int64')
+        return data, label
+
+    def __len__(self):
+        return len(self.data)
+
+
 class CustomModel(Module):
 
     def __init__(self):
@@ -79,12 +95,17 @@ def forward(self, x, foo=None):
 train_batch = 0
 test_batch = 0
 
+train_dataset = mnistdataset(data=X_train, label=y_train)
+train_loader = DataLoader(train_dataset, batch_size=batch_size, shuffle=True)
+val_dataset = mnistdataset(data=X_val, label=y_val)
+val_loader = DataLoader(val_dataset, batch_size=batch_size, shuffle=True)
+
 writer = SummaryWriter('runs/mlp')
 
 for epoch in range(n_epoch):  ## iterate the dataset n_epoch times
     start_time = time.time()
     ## iterate over the entire training set once (shuffle the data via training)
-    for X_batch, y_batch in tlx.utils.iterate.minibatches(X_train, y_train, batch_size, shuffle=True):
+    for X_batch, y_batch in train_loader:
         MLP.set_train()  # enable dropout
         with tf.GradientTape() as tape:
             ## compute outputs
@@ -98,7 +119,7 @@ def forward(self, x, foo=None):
     if epoch + 1 == 1 or (epoch + 1) % print_freq == 0:
         print("Epoch {} of {} took {}".format(epoch + 1, n_epoch, time.time() - start_time))
         train_loss, train_acc, n_iter = 0, 0, 0
-        for X_batch, y_batch in tlx.utils.iterate.minibatches(X_train, y_train, batch_size, shuffle=False):
+        for X_batch, y_batch in train_loader:
             train_batch += 1
             _logits = MLP(X_batch)
             train_loss += tlx.losses.softmax_cross_entropy_with_logits(_logits, y_batch)
@@ -112,7 +133,7 @@ def forward(self, x, foo=None):
             writer.add_scalar('train acc', train_acc / n_iter, train_batch)
 
         val_loss, val_acc, n_iter = 0, 0, 0
-        for X_batch, y_batch in tlx.utils.iterate.minibatches(X_val, y_val, batch_size, shuffle=False):
+        for X_batch, y_batch in val_loader:
             test_batch += 1
             _logits = MLP(X_batch)  # is_train=False, disable dropout
             val_loss += tlx.losses.softmax_cross_entropy_with_logits(_logits, y_batch)

examples/model_zoo/model/resnet50_weights_tf_dim_ordering_tf_kernels.h5

@@ -17,8 +17,8 @@
 resnet = ResNet50(pretrained=True)
 resnet.set_eval()
 
-img1 = tlx.vis.read_image('data/tiger.jpeg')
-img1 = tlx.prepro.imresize(img1, (224, 224))[:, :, ::-1]
+img1 = tlx.vision.load_image('data/tiger.jpeg')
+img1 = tlx.vision.transforms.Resize((224, 224))(img1)[:, :, ::-1]
 img1 = img1 - np.array([103.939, 116.779, 123.68]).reshape((1, 1, 3))
 
 img1 = img1.astype(np.float32)[np.newaxis, ...]

examples/model_zoo/pretrained_resnet50.py

@@ -17,8 +17,8 @@
 vgg = vgg16(pretrained=True)
 vgg.set_eval()
 
-img = tlx.vis.read_image('data/tiger.jpeg')
-img = tlx.prepro.imresize(img, (224, 224)).astype(np.float32) / 255
+img = tlx.vision.load_image('data/tiger.jpeg')
+img = tlx.vision.transforms.Resize((224, 224))(img).astype(np.float32) / 255
 
 start_time = time.time()
 output = vgg(img)

examples/model_zoo/pretrained_vgg16.py

@@ -21,10 +21,8 @@
     from tensorlayerx import losses
     from tensorlayerx import decorators
     from tensorlayerx import files
-    # from .utils import iterate
     from .utils import lazy_imports
     from . import utils
-    # from .text import nlp
     from tensorlayerx import logging
     from tensorlayerx import model
     from tensorlayerx import optimizers
@@ -34,20 +32,6 @@
 
     from tensorlayerx.utils.lazy_imports import LazyImport
 
-    # Lazy Imports
-    # db = LazyImport("tensorlayerx.utils.db")
-    # distributed = LazyImport("tensorlayerx.utils.distributed")
-    # nlp = LazyImport("tensorlayerx.text.nlp")
-    prepro = LazyImport("tensorlayerx.utils.prepro")
-    utils = LazyImport("tensorlayerx.utils")
-    visualize = LazyImport("tensorlayerx.utils.visualize")
-
-    # alias
-    vis = visualize
-
-    # alphas = array_ops.alphas
-    # alphas_like = array_ops.alphas_like
-
     # global vars
     global_flag = {}
     global_dict = {}

tensorlayerx/__init__.py

@@ -395,7 +395,7 @@ def ternary_operation(x):
 ########## Module Private Functions ##########
 
 
-@tf.RegisterGradient("TL_Sign_QuantizeGrad")
+# @tf.RegisterGradient("TL_Sign_QuantizeGrad")
 def _quantize_grad(op, grad):
     """Clip and binarize tensor using the straight through estimator (STE) for the gradient."""
     return tf.clip_by_value(grad, -1, 1)

tensorlayerx/nn/layers/utils.py

@@ -1,12 +1,7 @@
 #!/usr/bin/env python
 # -*- coding: utf-8 -*-
 
-# from .array_ops import *
-# from .db import *
-# from .distributed import *
 from .iterate import *
 from .prepro import *
-# from .rein import *
-# from .utils import *
-from .visualize import *
 from .lazy_imports import *
+from .visualize import *