All problems
Top-k Sampling Threshold
Problem
Return the top-k sampling threshold for a probability distribution: the k-th largest probability in probs.
Tokens with probability below this threshold are excluded by top-k sampling. Return 0.0 when k is not positive, k exceeds the number of probabilities, or probs is empty.
Hint
Sort descending and read the probability at position k - 1.
Test cases
case 1
in: [[0.5,0.3,0.2],2]
expected: 0.3
case 2
in: [[0.1,0.4,0.4,0.1],1]
expected: 0.4
case 3
in: [[0.25,0.25,0.25,0.25],3]
expected: 0.25
case 4
in: [[0.5,0.5],0]
expected: 0
case 5
in: [[0.5,0.5],5]
expected: 0
Starter codePython
def top_k_sampling_threshold(probs, k):
# Your code here
pass5 test cases · run your code against them in the editor