Pre-training and fine-tuning concept in Prompt Engineering / GenAI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to load a pre-trained model.

Prompt Engineering / GenAI

model = load_model('[1]')

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

Arandom_weights

Bpretrained_weights

Cuntrained_model

Dempty_model

Attempts:

3 left

💡 Hint

Common Mistakes

Choosing random or empty weights which means no prior learning.

Loading pretrained_weights means the model already learned from a large dataset.

2fill in blank

medium

Complete the code to fine-tune the model on new data.

Prompt Engineering / GenAI

model.train(new_data, epochs=[1])

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

A1000

B0

C-1

D5

Attempts:

3 left

💡 Hint

Common Mistakes

Using zero epochs means no training.

Using very large epochs may overfit the model.

Fine-tuning usually involves training for a few epochs like 5 to adjust the model gently.

3fill in blank

hard

Fix the error in the code to freeze pre-trained layers before fine-tuning.

Prompt Engineering / GenAI

for layer in model.layers:
    layer.[1] = False

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

Atrainable

Bfrozen

Clocked

Dfixed

Attempts:

3 left

💡 Hint

Common Mistakes

Using non-existent properties like 'frozen' or 'locked'.

Setting trainable = False freezes the layer so it is not updated during fine-tuning.

4fill in blank

hard

Fill both blanks to create a fine-tuning loop that updates only unfrozen layers.

Prompt Engineering / GenAI

for layer in model.layers:
    if layer.[1] == [2]:
        update(layer)

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

Atrainable

BTrue

CFalse

Dfrozen

Attempts:

3 left

💡 Hint

Common Mistakes

Checking for 'frozen' property which does not exist.

Using False instead of True to update layers.

We check if trainable == True to update only layers that can learn.

5fill in blank

hard

Fill all three blanks to create a dictionary comprehension that stores layer names and their trainable status.

Prompt Engineering / GenAI

layer_status = {layer.[1]: layer.[2] for layer in model.[3]

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

Aname

Btrainable

Clayers

Dweights

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'weights' instead of 'layers' to iterate.

Mixing up 'name' and 'trainable' properties.

This code creates a dictionary with layer names as keys and their trainable status as values from the model's layers.