{
    "byline": null,
    "dir": null,
    "excerpt": "They use the word-piece model from \u201cJapanese and Korean Voice Search\u201d, with 32,000 word pieces. (This is a lot less than the 200,000 used in that paper.) They state in the paper that the shared word-piece model is very similar to Byte-Pair-Encoding, which was used for NMT in this paper by researchers at U of Edinburgh.",
    "length": 1066,
    "siteName": null,
    "title": "Google's multilingual neural machine translation system"
}