model.fit() training loop in TensorFlow - Practice Problems & Coding Challenges

Challenge - 5 Problems

❓ Predict Output

intermediate

Output of model.fit() training metrics

What will be the output of the training metrics after running this code snippet?

TensorFlow

import tensorflow as tf
import numpy as np

# Simple dataset
x = np.array([[0.], [1.], [2.], [3.], [4.], [5.]], dtype=float)
y = np.array([[0.], [2.], [4.], [6.], [8.], [10.]], dtype=float)

# Simple linear model
model = tf.keras.Sequential([tf.keras.layers.Dense(1, input_shape=(1,))])
model.compile(optimizer='sgd', loss='mse', metrics=['mae'])

history = model.fit(x, y, epochs=3, verbose=0)

print(history.history)

A {"loss": [4.0, 0.25, 0.0625], "mae": [2.0, 1.0, 0.5]}

B {"loss": [4.0, 0.5, 0.125], "mae": [2.0, 1.0, 0.5]}

C {"loss": [4.0, 0.25, 0.0625], "mae": [1.5, 0.5, 0.25]}

D {"loss": [4.0, 0.5, 0.125], "mae": [1.5, 0.5, 0.25]}

💡 Hint

Look at how loss and mae decrease over epochs in a simple linear regression with SGD optimizer.

The mean squared error (loss) and mean absolute error (mae) decrease as the model learns. The values in option B match typical outputs for this simple training.

🧠 Conceptual

intermediate

Understanding batch size effect in model.fit()

Which statement correctly describes the effect of batch size in the model.fit() training loop?

A A larger batch size uses more memory but can lead to more stable gradient estimates during training.

B A smaller batch size always leads to faster training and better model accuracy.

C Batch size does not affect training speed or model performance.

D Batch size only affects the number of epochs, not the training steps per epoch.

💡 Hint

Think about how batch size influences memory use and gradient calculation.

Larger batch sizes use more memory but provide more stable gradient estimates, which can improve training stability. Smaller batches can be noisy but may generalize better.

❓ Hyperparameter

advanced

Choosing the right number of epochs in model.fit()

If a model is overfitting the training data during model.fit(), which adjustment is most appropriate?

A Decrease the number of epochs to stop training earlier.

B Increase the batch size to reduce overfitting.

C Increase the number of epochs to let the model learn more.

D Use a higher learning rate to speed up training.

💡 Hint

Overfitting means the model learns too much from training data and performs worse on new data.

Reducing epochs helps prevent the model from learning noise in the training data, which causes overfitting.

🔧 Debug

advanced

Identifying error in model.fit() usage

What error will this code raise when calling model.fit()?

TensorFlow

import tensorflow as tf

model = tf.keras.Sequential([tf.keras.layers.Dense(1)])
model.compile(optimizer='adam', loss='mse')

# Missing input data
model.fit(epochs=5)

A No error, training runs with default data

B ValueError: Data provided to model is empty

C RuntimeError: Model not compiled

D TypeError: fit() missing 1 required positional argument: 'x'

💡 Hint

Check the required arguments for model.fit()

model.fit() requires input data 'x' and target 'y' unless using a dataset. Omitting 'x' causes a TypeError.

❓ Model Choice

expert

Selecting model architecture for time series forecasting with model.fit()

You want to predict future values of a time series using model.fit(). Which model architecture is best suited for this task?

A A convolutional neural network (CNN) designed for image classification

B A simple feedforward neural network with Dense layers only

C A recurrent neural network (RNN) or LSTM model designed to handle sequences

D A clustering model like KMeans

💡 Hint

Time series data has order and depends on previous values.

RNNs and LSTMs are designed to process sequences and remember past information, making them ideal for time series forecasting.