Easy

DDPM Closed-Form Noising

~10 mincode completion

Implement q_sample(x0, eps, alpha_bar) returning of the same shape as x0.

Examples

eps=0, abar=0.5 scales x0 by sqrt(1/2)

Input
q_sample(1, 0, 0.5)
Output
0.70711

x0=0, abar=0.5 leaves sqrt(1/2) * eps

Input
q_sample(0, 2, 0.5)
Output
1.41421

A pair at abar=0.25

Input
q_sample([1, 0], [0, 1], 0.25)
Output
[0.5, 0.86603]

Hints

Hint 1

Take the square root at the end, not inside the sum.

Hint 2

Double check the order of the two square roots.

Requirements

  • alpha_bar: scalar in (0, 1]

  • Return xt, same shape as x0

Constraints

  • Allowed library: NumPy only

  • Time limit: 200 ms, Memory: 64 MB

Try similar problems(4)
Easy
GAN Discriminator BCE

Generative Image Models · ~10 min

Easy
KL of a Diagonal Gaussian to N(0, I)

Generative Image Models · ~12 min

Easy
Predict x0 from Noise

Generative Image Models · ~10 min

Easy
VAE Loss: Reconstruction plus KL

Generative Image Models · ~10 min

Where this shows up

~10 min

8 employers weight this skill

4 autonomy companies, 1 enterprise vendor, 1 defense company, 1 health and bio company, 1 AI product company. Top match scores 87.

Python
import numpy as np


def q_sample(x0, eps, alpha_bar):
    """
    Closed-form DDPM forward process.

    Args:
        x0, eps: arrays of the same shape
        alpha_bar: scalar in (0, 1]

    Returns:
        xt, same shape as x0
    """
    # YOUR CODE HERE
    pass
Loading docs…

The AI Mentor needs an account

It reads your code and the failing tests and nudges you toward the fix without handing you the answer. Free accounts get it on every problem you're working on today.