Self-improving agents in Agentic AI - Interactive Code Practice

Practice - 5 Tasks

1fill in blank

easy

Complete the code to define a self-improving agent class with an update method.

Agentic AI

class SelfImprovingAgent:
    def __init__(self, model):
        self.model = model

    def [1](self, data):
        # Improve the model using new data
        self.model.train(data)

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

Aevaluate

Bpredict

Cinitialize

Dupdate

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'predict' instead of 'update' because prediction does not improve the model.

Using 'initialize' which is for setup, not improvement.

The method to improve or refine the agent's model is typically called 'update'.

2fill in blank

medium

Complete the code to make the agent improve itself by retraining on its own predictions.

Agentic AI

def self_improve(agent, data):
    predictions = agent.predict(data)
    agent.[1](predictions)

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

Ainitialize

Breset

Cupdate

Devaluate

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'initialize' which resets the model instead of improving it.

Using 'evaluate' which only measures performance.

The agent should update its model using the predictions to improve itself.

3fill in blank

hard

Fix the error in the code to correctly implement a self-improving loop.

Agentic AI

for epoch in range(10):
    predictions = agent.predict(data)
    agent.[1](predictions)
    loss = agent.evaluate(data)
    print(f"Epoch {epoch}: Loss = {loss}")

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

Aupdate

Bpredict

Ctrain

Devaluate

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'predict' which does not change the model.

Using 'evaluate' which only measures performance.

The agent should update its model with predictions to improve, not predict again.

4fill in blank

hard

Fill both blanks to create a dictionary comprehension that maps data points to their improved predictions.

Agentic AI

improved_predictions = {point: agent.[1](point) for point in data if agent.[2](point) > 0.5}

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

Apredict

Bevaluate

Cconfidence_score

Dupdate

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'update' in the comprehension which changes the model instead of predicting.

Using 'evaluate' which returns overall loss, not per data point score.

We predict for each point and filter by confidence_score greater than 0.5.

5fill in blank

hard

Fill all three blanks to implement a self-improving agent loop that predicts, filters, and updates.

Agentic AI

for data_point in dataset:
    prediction = agent.[1](data_point)
    if prediction [2] 0.7:
        agent.[3]([data_point])

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

Apredict

B>

Cupdate

Devaluate

Attempts:

3 left

💡 Hint

Common Mistakes

Using 'evaluate' instead of 'update' to improve the model.

Using '<' instead of '>' which reverses the condition.

The agent predicts, checks if prediction is greater than 0.7, then updates with that data point.