Added instructions of how to visualize tensorflow in some examples and

generalized gradients for TF

Author: ethantang95

docs/DockerInstall.md

@@ -10,20 +10,20 @@ To run DIGITS smoothly, Nvidia Docker must be installed. Follow [this guide](htt
 To fetch the latest DIGITS's docker image, run the following:
 
 ```bash
-sudo docker pull nvidia/digits:latest
+docker pull nvidia/digits:latest
 ```
 
 To fetch DIGITS 6 RC, run the following:
 
 ```bash
-sudo docker pull nvidia/digits.6-rc
+docker pull nvidia/digits.6-rc
 ```
 
 ## Running DIGITS on Nvidia Docker
 To run DIGITS with Nvidia Docker, use the following command:
 
 ```bash
-sudo docker run -v <path to data>:/data/ -p 5000:5000 nvidia/digits:latest
+nvidia-docker run -v <path to data>:/data/ -p 5000:5000 nvidia/digits:latest
 ```
 
 where `<path to data>` is where you have stored data sets or other files necessary for DIGITS to use on your system.

examples/autoencoder/README.md

@@ -7,6 +7,7 @@ Table of Contents
 * [Model definition (Torch)](#model-creation-torch)
 * [Model definition (Tensorflow)](#model-creation-tensorflow)
 * [Verification](#verification)
+* [Visualization for Tensorflow](#visualization-for-tensorflow)
 
 ## Introduction
 
@@ -152,4 +153,10 @@ The reconstructed image, which looks similar to the normalized image:
 
 ![Reconstructed Image](reconstructed-image.png)
 
+### Visualization for Tensorflow
 
+To visualize the result of the autoencoder in Tensorflow properly, select `Image output` as `Visualization method`. Then select `HWC` as the `Data order`.
+
+![Tensorflow Settings](tensorflow-settings.png)
+
+Then follow the instructions for [verification](#verification)

examples/autoencoder/tensorflow-settings.png

@@ -105,7 +105,7 @@ The `create_images.py` script created many images we can use for testing.
 You may also run the command again to produce new images.
 To help visualize the output of the network, select the `Image` visualization method and set `Pixel Conversion` to `clip`:
 
-For Tensorflow only, select `HWC` for data order
+**For Tensorflow only, select `HWC` for data order**
 
 ![select visualization](select-visualization.png)
 

examples/binary-segmentation/README.md

@@ -160,17 +160,19 @@ class UserModel(Tower):
 
     @model_property
     def inference(self):
+        dimension_h = self.input_shape[0]
+        dimension_w = self.input_shape[1]
         const = tf.constant(0.004)
         normed = tf.multiply(self.x, const)
 
         # The reshaping have to be done for tensorflow to get the shape right
-        right_shape = tf.reshape(normed, shape=[-1, 50, 50]) 
+        right_shape = tf.reshape(normed, shape=[-1, dimension_h, dimension_w]) 
         transposed = tf.transpose(right_shape, [0, 2, 1])
-        squeezed = tf.reshape(transposed, shape=[-1, 2500])
+        squeezed = tf.reshape(transposed, shape=[-1, dimension_h * dimension_w])
 
         # Define weights
         weights = {
-            'w1': tf.get_variable('w1', [2500, 2])
+            'w1': tf.get_variable('w1', [dimension_h * dimension_w, 2])
         }
         biases = {
             'b1': tf.get_variable('b1', [2])

examples/regression/README.md

@@ -47,13 +47,13 @@ class DatasetForm(Form):
 
     image_width = wtforms.IntegerField(
         u'Image Width',
-        default=32,
+        default=50,
         validators=[validators.DataRequired()]
     )
 
     image_height = wtforms.IntegerField(
         u'Image Height',
-        default=32,
+        default=50,
         validators=[validators.DataRequired()]
     )