Diffusion MSE Loss

Problem

Compute the diffusion training loss: mean squared error between the true noise and the predicted noise, mean((eps - eps_pred)^2).

The signature is diffusion_mse_loss(eps, eps_pred).

Hint

Predicting the noise is equivalent to predicting the score, up to scaling.

case 1

in: [[1,2],[1,2]]

expected: 0

case 2

in: [[1,2],[2,4]]

expected: 2.5

case 3

in: [[0.5],[1]]

expected: 0.25

case 4

in: [[-1,0,1],[0,0,0]]

expected: 0.6666666667

Starter codePython

def diffusion_mse_loss(eps, eps_pred):
    # Your code here
    pass

4 test cases · run your code against them in the editor