N-Step Q Target

Problem

Compute the n-step Q-learning target from a list of n rewards.

G = sum of gamma^t * rewards[t]. If the episode is not done, add gamma^n * next_q; if it is done, do not bootstrap.

Hint

The discount applied to next_q is gamma raised to the number of collected rewards.

case 1

in: [[1,1],0.9,5,false]

expected: 5.950000000000001

case 2

in: [[1,1],0.9,100,true]

expected: 1.9

case 3

in: [[],0.9,3,false]

expected: 3

case 4

in: [[2],0.5,4,false]

expected: 4

Starter codePython

def n_step_q_target(rewards, gamma, next_q, done):
    # Your code here
    pass

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