Understanding beam search


I’m trying to understand beam search. I have generated an HTML output and have looked at the tree for a sentence. The different branches could be more or less good translations. Can I tell from the numbers on the nodes why a particular translation was chosen?

Best regards


Yes. Higher score means the hypothesis is better.

If you want an in-depth understanding of the beam search, I think you can easily find learning materials on Google.

Thank you for the answer. Yes I add all the scores at every branch and the highest score was the translation.