MediumNLPnlp-154

NLI Entailment Logits

Problem

Compute NLI entailment logits. Build pair features as the element-wise product of premise and hypothesis, then for each class row return dot(row, features) + bias[class].

class_vectors has one row per class and bias has one entry per class, both returned in input order.

Hint

The interaction features are products; each class is then a linear layer over them.

Test cases

case 1

in: [[1,0],[1,1],[[1,0],[0,1],[0.5,0.5]],[0.1,-0.1,0]]

expected: [1.1,-0.1,0.5]

case 2

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

expected: [0.5]

case 3

in: [[1,2],[0,0],[[1,1],[1,-1]],[0,0]]

expected: [0,0]

case 4

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

expected: [0.5]

Starter codePython

def nli_entailment_logits(premise, hypothesis, class_vectors, bias):
    # Your code here
    pass

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