AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Untangling tradeoffs between recurrence and self-attention in artificial neural networks

Giancarlo Kerg, Bhargav Kanuparthi, Anirudh Goyal ALIAS PARTH GOYAL, Kyle Goyette, Yoshua Bengio, Guillaume Lajoie

NeurIPS 2020 · Architectures and efficiency › Attention and transformers · approach: Sparse and structured attention · 6 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 relevancy screening (RelRNN vs. sparse attention SAB without screening)RNN with sparse attention · Transfer Copy T=5000accuracy (%): 95 -> 99Improved
add relevancy-screened attention to LSTM (RelLSTM vs LSTM)LSTM · Transfer Copy T=5000accuracy (%): 14 -> 99Improved
add relevancy screening to RNN (RelRNN vs expRNN)RNN · Transfer Copy T=5000accuracy (%): 50 -> 99Improved
use full self-attention (MemRNN) instead of screened sparse attentionRelRNN · Transfer Copy T=5000accuracy (%): 99 -> OOMHurt
add relevancy screening (RelLSTM vs LSTM)LSTM · Denoise T=2000accuracy (%): 15 -> 99Improved
add relevancy screening to RNN (RelRNN vs orth-RNN)RNN · Denoise T=2000accuracy (%): 3 -> 99Improved
use full attention MemRNN instead of RelRNN at long TRelRNN · Denoise T=2000accuracy (%): 99 -> OOMHurt
add relevancy screening to RNN (RelRNN vs RNN)RNN · PTB char-levelBPC: 1.56 -> 1.43Improved
add relevancy screening to RNN (RelRNN vs RNN)RNN · PTB char-levelaccuracy (%): 66 -> 69Improved
add relevancy screening to LSTM (RelLSTM vs LSTM)LSTM · PTB char-levelBPC: 1.36 -> 1.36No clear effect
add relevancy screening to LSTM (RelLSTM vs LSTM)LSTM · pMNISTaccuracy (%): 91.1 -> 94.3Improved
add relevancy screening to RNN (RelRNN vs expRNN)RNN · pMNISTaccuracy (%): 96.6 -> 92.8Hurt

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