Damerau-Levenshtein distance or Levenshtein distance for LTA

Hi, for the long-term anticipation task, the paper says Damerau-Levenshtein distance is used for evaluation but the official code uses editdistance package, where only the Levenshtein distance is implemented, with transposition operation missing.

>>> import editdistance
>>> editdistance.eval([1, 2], [2, 1])

So for the online evaluation, which edit distance can we assume will be used there?

1 Like

The evaluation server also uses Levenshtein distance from the editdistance package.

1 Like