2026-10-14T14:53:06Z https://www.scidb.cn/oai

10.57760/sciencedb.j00104.00151 2026-09-10T10:52:15Z

2026-09-10 IWSLT14 German-English, WMT14 English-German, and WMT16 English-Romanian doi:10.57760/sciencedb.j00104.00151 en We conducted experiments on three widely used machine translation benchmarks: IWSLT14 German-English (De-En, ~160K sentence pairs), WMT14 English-German (En-De, ~4.5M sentence pairs), and WMT16 English-Romanian (En-Ro, ~610K sentence pairs). All datasets follow the standard splits of the respective benchmarks, and subword vocabularies are constructed using Byte Pair Encoding (BPE) to alleviate vocabulary sparsity and unify token granularity. IWSLT14; WMT14; WMT16 Mengyao,Wang PUBLIC https://creativecommons.org/licenses/by/4.0/ dataset Science Data Bank