Retraining strategies in ML Python - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to retrain a model using new data.

ML Python

model.fit(new_data, new_labels, epochs=[1])

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

A5

B-1

C0

D10

Attempts:

3 left

💡 Hint

Common Mistakes

Using 0 epochs means no training happens.

Negative epochs cause errors.

Setting epochs to 10 allows the model to train for 10 cycles on the new data, which is a common choice for retraining.

2fill in blank

medium

Complete the code to freeze all layers except the last one before retraining.

ML Python

for layer in model.layers[:-1]:
    layer.trainable = [1]

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

AFalse

BTrue

CNone

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Setting trainable to True retrains all layers.

Using None or 0 is invalid for trainable property.

Setting trainable to False freezes the layers so they are not updated during retraining.

3fill in blank

hard

Fix the error in the code to compile the model with a suitable optimizer for retraining.

ML Python

model.compile(optimizer=[1], loss='categorical_crossentropy', metrics=['accuracy'])

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

A'sgd'

B'adagrad'

C'adam'

D'rmsprop'

Attempts:

3 left

💡 Hint

Common Mistakes

Using unsupported optimizer names causes errors.

Choosing SGD may require tuning learning rate.

Adam optimizer is widely used for retraining because it adapts learning rates and converges well.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that filters new data samples with label 1 for retraining.

ML Python

filtered_data = {i: x for i, x in enumerate(new_data) if new_labels[i] [1] [2]

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

A==

B!=

C1

D0

Attempts:

3 left

💡 Hint

Common Mistakes

Using '!=' filters out the desired class.

Using 0 instead of 1 filters the wrong class.

The comprehension keeps samples where the label equals 1, which is useful to retrain on a specific class.

5fill in blank

hard

Fill all three blanks to update the learning rate and retrain the model with early stopping.

ML Python

from tensorflow.keras.callbacks import EarlyStopping

optimizer = tf.keras.optimizers.Adam(learning_rate=[1])
model.compile(optimizer=optimizer, loss='categorical_crossentropy', metrics=['accuracy'])
early_stop = EarlyStopping(monitor='val_loss', patience=[2], restore_best_weights=[3])
model.fit(train_data, train_labels, epochs=20, validation_split=0.2, callbacks=[early_stop])

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

A0.001

B3

CTrue

D0.1

Attempts:

3 left

💡 Hint

Common Mistakes

Using too high learning rate causes training instability.

Setting patience too low stops training too early.

Not restoring best weights may keep worse model.

Learning rate 0.001 is common for Adam optimizer. Patience 3 waits 3 epochs without improvement before stopping. Restoring best weights ensures best model is kept.