EasyNLPnlp-089

Subword Reconstruction

Problem

Reconstruct words from subword tokens. A token starting with '##' continues the previous word, and any other token starts a new word.

Return the list of reconstructed words.

Hint

Accumulate continuation pieces into the current word and flush before starting a new one.

case 1

in: [["play","##ing"]]

expected: ["playing"]

case 2

in: [["un","##want","##ed"]]

expected: ["unwanted"]

case 3

in: [["hello","world"]]

expected: ["hello","world"]

case 4

in: [["##ed"]]

expected: ["ed"]

case 5

in: [[]]

expected: []

Starter codePython

def subword_reconstruction(tokens):
    # Your code here
    pass

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