How to Cast Integer to Float and Vice Versa in TensorFlow

In this article, we will learn to convert the data types of tensors to integer and float.

First, let's create two nodes.
node1 = tf.constant(5)    # dtype is int32
node2 = tf.constant(6.0)  # dtype is float32
Let's multiply the two nodes to create a new node
node3 = tf.multiply(node1, node2)
This will throw a
TypeError: Input 'y' of 'Mul' Op has type float32 that does not match type
int32 of argument 'x'.
This is because the node1 data type is an integer while the data type of node2 is a float.

Multiply method doesn't convert the types of its variables automatically. For now, we need to change them before performing any type of arithmetic operations on them. I hope this changes in the future releases.

The correct way is to convert either of the node's type to match the other. Here, we are converting the node1 which is an integer to a float.
tf.to_float(node1, name='ToFloat')
Replacing the node1 with the above line in the multiply method will resolve the issue. To learn, more about other types of casting you can check the official website.

Below is a full code that correctly performs multiplication of two nodes when there are of different types.

import tensorflow as tf
node1 = tf.constant(5) # dtype is int32
node2 = tf.constant(6.0) # dtype is float32
print('Node 1: ', node1) # Node 1: Tensor("Const:0", shape=(), dtype=int32)
print('Node 2: ', node2) # Node 2: Tensor("Const_1:0", shape=(), dtype=float32)
# Multiplication
# Convert the node1 type to float32 before multiplying
# If you don't convert the types, you'll get TypeError: Input 'y' of 'Mul' Op has type float32 that does not match type int32 of argument 'x'
node3 = tf.multiply(tf.to_float(node1, name='ToFloat'), node2) # dtype is float32
# Convert the node2 type to int32 before multiplying
node4 = tf.multiply(node1, tf.to_int32(node2, name='ToInt32')) # dtype is int32
print('Node3: Multiply: ', node3) # Node3: Multiply: Tensor("Mul:0", shape=(), dtype=float32)
print('Node4: Multiply: ', node4) # Node4: Multiply: Tensor("Mul_1:0", shape=(), dtype=int32)
# Run inside a session to get the values
with tf.Session() as sess:
node3Value = sess.run(node3)
node4Value = sess.run(node4)
print('Multiplication value at node 3: ', node3Value) # Multiplication value at node 3: 30.0
print('Multiplication value at node 4: ', node4Value) # Multiplication value at node 4: 30
view raw Solution to TypeError or ValueError caused by Tensor dtype hosted with ❤ by GitHub
In a different scenario, you'll get an error
ValueError: Tensor conversion requested dtype int32 for Tensor with dtype
float64 in Tensorflow.
The solution is same. Convert the first tensor or the second tensor type to match the other.

Popular posts from this blog

Create Assets Folder, Add Files and Read Data From It

Image
Sometimes, we need to read the data stored in the files during runtime and want to make sure the data is not tampered by the Android build system. In this article, we will take a look at how to create the assets folder using Android Studio and read the stored data from it. First, Right-click on the module name or the package name. This opens an option window where we need to select the Assets Folder option by hovering over New and then on the Folder option. Now, we will see a window where we change the folder location and the source set. Change the values if necessary otherwise keep the defaults and click Finish. It may trigger a Gradle Sync. Irrespective of the sync, the Assets folder is created and is visible in the module. Now, Right-click on the Assets folder and click on the File option by hovering over the New option. Enter the name of the file name and click Enter to create the file. The file with the name user_details.json is created under the assets folder. Add some JSON...
Read more

Run the Emulator directly in Android Studio

Image
In Android Studio 4.1, a minor feature is released to reduce the screen (or window) switching between Android Studio and the Emulator. Using this feature, we can now run the Emulator directly inside the Android Studio. We can make the changes to the code and see the changes reflect in the emulator in the same window similar to layout XML and its preview. To achieve this, open the Preferences window and select the Emulator menu item under the Tools expandable menu. Here we can check the option Launch in a tool window and click OK . Now run the Emulator to see it inside the Android Studio window.
Read more

How to Read Metadata from AndriodManifest File

Metadata tags are key-value pairs and one of the decent ways to access the API keys from the libraries. They are added to the AndroidManifest.xml under the application tag and are available through the PackageManager . If you have worked with third-party libraries or even with App shortcuts, Downloadable Fonts, Firebase MLKit, you may have come across metadata tags. In the article, let’s add the metadata tags to the app module’s Manifest file and read the metadata key-values and the key-resource from a library module both in kotlin and java. Add meta-data: Add metadata to the app module's AndroidManifest.xml file under the application tag. You can add as many metadata elements as you want and you can also add metadata under the activity, service, broadcast receiver, content provider tags. <meta-data android:name= "com.gsrikar.library.LIBRARY_ACCESS_KEY" android:value= "masndfiu4er9funwikesd" /> <meta-data and...
Read more

Add Spacing to Recycler View Linear Layout Manager Using Item Decoration

Image
In this article we will learn about, item decoration in Android and how we can use it to add even spacing between the child items. A RecyclerView without spacing will look like the image below but in most cases, we want to add spacing between the items. [You can see there are no margins between the views and only elevation] A more traditional option is adding margins in the list item xml file. If we take this route, we will soon end up with uneven spacing issues with vertical margins. [There are even left, right, top, bottom margins for all the rows except the last item in the first image and in the subsequent image, the top margin for the first item is half of the other items] The best way is to add the ItemDecoration to the RecyclerView as it affects the individual items and you can have more control over them. Kotlin: /** * Add vertical, horizontal spacing to the recycler view */ class DefaultItemDecorator(private val horizontalSpacing: Int, ...
Read more

How to Set an Android App as a Default Dialer

Image
A default dialer is an application that handles the Phone calls on the device. Google made the provision for third-party applications to be a default phone handler from SDK 23 (6.0, Marshmallow version). In the article, we will learn about how to set our Android application as a default dialer. 1) Add the below intent filters under the activity tag in the AndroidManifest.xml. <intent-filter> <action android:name="android.intent.action.DIAL"/> <category android:name="android.intent.category.DEFAULT"/> </intent-filter> <intent-filter> <action android:name="android.intent.action.DIAL"/> <category android:name="android.intent.category.DEFAULT"/> <data android:scheme="tel"/> </intent-filter> 2) Check if our application is already a default dialer app using TelecomManager. val telecomManager = getSystemService(Context.TELECOM_SERVICE) as TelecomManager Log...
Read more