AI Paper Atlas

Research map / Image generation / Autoregressive and latent generation

Modeling Coherence for Discourse Neural Machine Translation

Hao Xiong, Zhongjun He, Hua Wu, Haifeng Wang

AAAI 2019 · Image generation › Autoregressive and latent generation · approach: Synthetic training data from generators

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Add second-pass document deliberationfirst-pass Transformer · IWSLT 2015 Zh-En TED talks (test average)BLEU: 20.41 -> 20.92Improved
Add reward-teacher policy learningtwo-pass Transformer · IWSLT 2015 Zh-En TED talks (test average)BLEU_doc: 26.89 -> 27.50Improved
Add reward-teacher policy learningfirst-pass Transformer · IWSLT 2015 Zh-En TED talks (test average)BLEU_doc: 25.99 -> 26.82Improved
Add discourse BLEU as an additional rewardtwo-pass-rl Transformer · IWSLT 2015 Zh-En TED talks (test average)BLEU_doc: +2.2 BLEUImproved
Train with talks in original sentence order rather than randomly shuffled sentencesTransformer training · IWSLT 2015 Zh-En TED talks (test average)BLEU: 20.10 -> 20.41Improved
Pretrain on WMT2018 bilingual data before TED fine-tuningtwo-pass-bleu-rl · IWSLT 2015 Zh-En TED talks (test average)BLEU: 21.33 -> 27.55Improved
Replace TED reward-teacher training corpus with AFP (ablation)two-pass-rl · IWSLT 2015 Zh-En TED talks (test average)BLEU: 21.11 -> 20.54Hurt
Replace TED reward-teacher training corpus with XIN (ablation)two-pass-rl · IWSLT 2015 Zh-En TED talks (test average)BLEU: 21.11 -> 20.92Hurt
Replace TED reward-teacher training corpus with Gigaword (ablation)two-pass-rl · IWSLT 2015 Zh-En TED talks (test average)BLEU: 21.11 -> 20.49Hurt
Set balance factors to λ1=0.85 and λ2=0.80 (ablation)first-pass-rl and two-pass-rl · IWSLT 2015 Zh-En TED talksBLEUImproved
Give the second pass context from up to 16 surrounding sentences rather than one preceding sentencetwo-pass versus context-encoder · IWSLT 2015 Zh-En TED talks (test average)BLEU: 20.31 -> 20.92Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy