GRU layer in TensorFlow - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to create a GRU layer with 32 units.

TensorFlow

gru_layer = tf.keras.layers.GRU([1])

Drag options to blanks, or click blank then click option'

A32

B64

C16

D128

Attempts:

3 left

💡 Hint

Common Mistakes

Using a number too large or too small without reason.

Forgetting to specify the number of units.

The GRU layer is created with 32 units, which controls the output dimensionality.

2fill in blank

medium

Complete the code to add a GRU layer that returns sequences.

TensorFlow

gru_layer = tf.keras.layers.GRU(64, [1]=True)

Drag options to blanks, or click blank then click option'

Areturn_state

Bgo_backwards

Cstateful

Dreturn_sequences

Attempts:

3 left

💡 Hint

Common Mistakes

Using return_state instead, which returns the last state and hidden state separately.

Forgetting to set this argument when sequence output is needed.

The return_sequences argument makes the GRU layer output the full sequence of hidden states.

3fill in blank

hard

Fix the error in the GRU layer creation by completing the missing argument.

TensorFlow

gru_layer = tf.keras.layers.GRU(128, activation=[1])

Drag options to blanks, or click blank then click option'

A'softmax'

B'sigmoid'

C'tanh'

D'relu'

Attempts:

3 left

💡 Hint

Common Mistakes

Using softmax which is for classification outputs.

Using sigmoid which is used inside gates but not as main activation.

The GRU layer typically uses the tanh activation function for its internal state updates.

4fill in blank

hard

Fill both blanks to create a GRU layer with 50 units and dropout rate of 0.2.

TensorFlow

gru_layer = tf.keras.layers.GRU([1], dropout=[2])

Drag options to blanks, or click blank then click option'

A50

B0.2

C0.5

D100

Attempts:

3 left

💡 Hint

Common Mistakes

Confusing dropout with recurrent_dropout.

Using dropout values greater than 1.

The GRU layer is set with 50 units and a dropout rate of 0.2 to reduce overfitting.

5fill in blank

hard

Fill all three blanks to build a GRU layer with 128 units, returning sequences, and using 'relu' activation.

TensorFlow

gru_layer = tf.keras.layers.GRU([1], [2]=True, activation=[3])

Drag options to blanks, or click blank then click option'

A128

Breturn_sequences

C'relu'

D'tanh'

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'tanh' activation instead of 'relu' as requested.

Forgetting to set return_sequences to True.

This GRU layer has 128 units, returns the full sequence, and uses the 'relu' activation function.