AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Navigating with Graph Representations for Fast and Scalable Decoding of Neural Language Models

Minjia Zhang, Wenhan Wang, Xiaodong Liu, Jianfeng Gao, Yuxiong He

NeurIPS 2018 · Architectures and efficiency › RNNs and normalization · approach: Non-autoregressive translation · 9 citations (OpenAlex)

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
Replace full softmax with FGD at efSearch 50NMT decoder · IWSLT’14 German-Englishsoftmax execution time: 6.3ms -> 0.43msImproved
Replace full softmax with FGD at efSearch 20NMT decoder · IWSLT’14 German-EnglishBLEU: losing 0.67 BLEU scoreHurt
Increase efSearch from 20 to 50 (ablation)FGD · IWSLT’14 German-EnglishBLEU: 26.69 -> 29.06Improved
Increase efSearch from 20 to 50 (ablation)FGD · IWSLT’14 German-Englishsoftmax execution time: 0.21ms -> 0.43msHurt
Increase efSearch from 20 to 200 (ablation)FGD · IWSLT’14 German-Englishprecision@1: 0.939 -> 0.992Improved
Increase efSearch from 20 to 200 (ablation)FGD · IWSLT’14 German-Englishdistance calculations: 981 / 50K -> 5785 / 50KHurt
Increase beam size from 1 to 10 (ablation)FGD · IWSLT’14 German-EnglishBLEU: 26.69 -> 27.62 (efSearch 20); 27.55 -> 28.9 (efSearch 50)Mixed
Increase vocabulary size from 10K to 80K (ablation)FGD language model · WikiText-2precision@1: 0.938 -> 0.933 (efSearch 50); 1.000 -> 0.982 (efSearch 200)Mixed
Increase vocabulary size (ablation)FGD language model · WikiText-2decoding timeHurt
Use SVD-softmax instead of FGDNMT decoder · IWSLT’14 German-Englishsoftmax execution time: 0.43ms vs 5.53msHurt

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