SOTAVerified

Edinburgh Neural Machine Translation Systems for WMT 16

2016-06-09WS 2016Code Available0· sign in to hype

Rico Sennrich, Barry Haddow, Alexandra Birch

Code Available — Be the first to reproduce this paper.

Reproduce

Code

Abstract

We participated in the WMT 2016 shared news translation task by building neural translation systems for four language pairs, each trained in both directions: English<->Czech, English<->German, English<->Romanian and English<->Russian. Our systems are based on an attentional encoder-decoder, using BPE subword segmentation for open-vocabulary translation with a fixed vocabulary. We experimented with using automatic back-translations of the monolingual News corpus as additional training data, pervasive dropout, and target-bidirectional models. All reported methods give substantial improvements, and we see improvements of 4.3--11.2 BLEU over our baseline systems. In the human evaluation, our systems were the (tied) best constrained system for 7 out of 8 translation directions in which we participated.

Tasks

Benchmark Results

DatasetModelMetricClaimedVerifiedStatus
WMT2016 Czech-EnglishAttentional encoder-decoder + BPEBLEU score31.4Unverified
WMT2016 English-CzechAttentional encoder-decoder + BPEBLEU score25.8Unverified
WMT2016 English-GermanAttentional encoder-decoder + BPEBLEU score34.2Unverified
WMT2016 English-RomanianBiGRUBLEU score28.1Unverified
WMT2016 English-RussianAttentional encoder-decoder + BPEBLEU score26Unverified
WMT2016 German-EnglishAttentional encoder-decoder + BPEBLEU score38.6Unverified
WMT2016 Romanian-EnglishAttentional encoder-decoder + BPEBLEU score33.3Unverified
WMT2016 Russian-EnglishAttentional encoder-decoder + BPEBLEU score28Unverified

Reproductions