BEiT Tokenizer Target

Problem

Map a patch's RGB values to the nearest codebook entry by squared Euclidean distance and return its index, breaking ties by the smaller index.

The signature is beit_tokenizer_target(patch_rgb, codebook). Return the codebook index.

Hint

BEiT predicts discrete visual tokens from a frozen dVAE tokenizer instead of raw pixels.

case 1

in: [[1,0],[[1,0],[0,1],[0.5,0.5]]]

expected: 0

case 2

in: [[0,1],[[1,0],[0,1],[0.5,0.5]]]

expected: 1

case 3

in: [[0.4,0.6],[[1,0],[0,1],[0.5,0.5]]]

expected: 2

case 4

in: [[0.6,0.4],[[1,0],[0,1]]]

expected: 0

Starter codePython

def beit_tokenizer_target(patch_rgb, codebook):
    # Your code here
    pass

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