Type casting in TensorFlow - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

TensorFlow Type Casting Output

What is the output of this TensorFlow code snippet?

TensorFlow

import tensorflow as tf
x = tf.constant([1.7, 2.3, 3.9])
y = tf.cast(x, tf.int32)
print(y.numpy())

A [1 3 4]

B [2 3 4]

C [1 2 3]

D [2 2 3]

💡 Hint

Casting from float to int truncates the decimal part.

Casting floats to integers in TensorFlow truncates the decimal part, so 1.7 becomes 1, 2.3 becomes 2, and 3.9 becomes 3.

❓ Model Choice

intermediate

Choosing Correct TensorFlow Data Type for Image Pixels

You have image pixel values ranging from 0 to 255. Which TensorFlow data type is best to store these pixels efficiently without losing information?

A tf.uint8

B tf.float32

C tf.int32

D tf.bool

💡 Hint

Pixels are whole numbers between 0 and 255.

tf.uint8 stores unsigned 8-bit integers from 0 to 255, perfect for pixel values without extra memory usage.

🔧 Debug

advanced

Debugging TensorFlow Type Casting Error

What error will this TensorFlow code raise?

TensorFlow

import tensorflow as tf
x = tf.constant(['1', '2', '3'])
y = tf.cast(x, tf.int32)

A AttributeError: 'Tensor' object has no attribute 'astype'

B TypeError: Cannot cast string to int32

C No error, output: [1 2 3]

D ValueError: invalid literal for int() with base 10

💡 Hint

Casting strings directly to integers is not supported in TensorFlow.

TensorFlow cannot cast string tensors directly to integer tensors, so it raises a TypeError.

❓ Metrics

advanced

Effect of Type Casting on Model Accuracy

You cast your model's output tensor from float64 to float32 before calculating accuracy. What is the most likely effect?

A Model training stops with a type mismatch error

B Accuracy improves significantly due to faster computation

C Accuracy drops to zero due to data corruption

D Accuracy remains almost the same with minor precision loss

💡 Hint

Float32 has less precision than float64 but is usually enough for accuracy calculations.

Casting from float64 to float32 may cause minor precision loss but usually does not significantly affect accuracy.

🧠 Conceptual

expert

Why Use tf.cast in TensorFlow Pipelines?

Which is the best reason to use tf.cast in a TensorFlow data pipeline?

A To convert data types for compatibility with model input requirements

B To automatically normalize data values between 0 and 1

C To increase the size of tensors for better model performance

D To change tensor shapes dynamically during training

💡 Hint

Models expect inputs in specific data types.

tf.cast is used to convert tensors to the correct data type so the model can process them without errors.