AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Text Compression-aided Transformer Encoding

Zuchao Li, Zhuosheng Zhang, Hai Zhao, Rui Wang, Kehai Chen, Masao Utiyama, Eiichiro Sumita

TPAMI 2021 · Architectures and efficiency › Attention and transformers · approach: Efficient language transformers · 50 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
add ETC Pipeline backbone fusion (BBF)Transformer (base) · WMT14 EN-DEBLEU: 27.24 -> 28.35Improved
add ITC Joint (BBF) instead of pipeline ETC (ablation)Transformer (base) · WMT14 EN-DEBLEU: 28.35 -> 28.57Improved
fuse backbone on decoder side (BDF) instead of encoder side (BEF) (ablation)ETC Pipeline, Transformer (base) · WMT14 EN-DEBLEU: 27.75 -> 28.14Improved
use meaningful compression (Semi-supervised) instead of random sampling (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 26.53 -> 28.35Improved
add AllText as extra source instead of compression (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 27.24 -> 27.24No clear effect
use F8W (first eight words) as compression (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 27.24 -> 27.40No clear effect
use unsupervised instead of supervised compression (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 27.80 -> 27.97Improved
use semi-supervised finetuning instead of unsupervised (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 27.97 -> 28.35Improved
use independent encoders instead of shared encoder (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 28.35 -> 28.50Mixed
add DAE pre-training instead of compression (ablation)Transformer (base) · WMT14 EN-DEBLEU: 27.24 -> 27.65Improved
train ETC jointly (ETC Joint) instead of pipeline (ablation)ETC Pipeline BBF · WMT14 EN-DEBLEU: 28.35 -> 28.04Hurt
use ETC trained on matching domain (TED) (ablation)ETC Pipeline BBF · IWSLT15 TED testBLEU: 30.76 -> 31.28Improved

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