Does its embedding process runs on other languages(Ex. Korean)?

[andymogul]

I wonder if this seq2seq can do embedding on Korean language.

If it can't, how can I give input as embedded vector? (For example, give input as [0.9, 0.05, 0.05, 0, ...] instead of 'apple')

Above question means how can I use seq2seq2 model without embedding step

[ruohoruotsi]

I think there are some clues in this issue: https://github.com/google/seq2seq/issues/288

• Create a vocabulary
• Index them (word 2 Index)
• If you have a language embedding word2vec model like glove , assign each vector to word.
• Other wise let the words learn their vector while training