3

I've been looking at how BLEU score works. What I understood from the online videos + the original research paper is that BLEU score value should be within the range 0-1.

Then, when I started to look at some research papers, I found that BLEU value (almost) always higher than 1! For instance, have a look here:

Am I missing something?

Another small point: what does the headers in the table below mean? The BLEU score was calculated using unigrams, then unigrams & bigrams (averaged), etc.? or each ngrams size was calculated independently?

enter image description here

Minions
  • 5,104
  • 5
  • 50
  • 91
  • 2
    I believe the BLEU score in the table you show is the number (0 to 1) as you point out, but expressed as a percentage. See this ref: https://cloud.google.com/translate/automl/docs/evaluate – AJR Oct 06 '21 at 09:38
  • Can't be because in one definition human level in 100% while in another definition human level is 50-60. – Dims Jan 09 '23 at 11:57

0 Answers0