Behavior Cloning Dataset Split

Problem

Split n demonstration episodes into train and validation sets.

The validation size is int(n * val_fraction) using truncation, and the training set gets the rest. Return [n_train, n_val].

Hint

Truncation keeps the split deterministic for odd episode counts.

case 1

in: [10,0.2]

expected: [8,2]

case 2

in: [7,0.3]

expected: [5,2]

case 3

in: [5,0]

expected: [5,0]

case 4

in: [3,1]

expected: [0,3]

Starter codePython

def bc_dataset_split(n_episodes, val_fraction):
    # Your code here
    pass

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