AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Discrete-Valued Neural Communication

Dianbo Liu, Alex Lamb, Kenji Kawaguchi, Anirudh Goyal, Chen Sun, Michael Curtis Mozer, Yoshua Bengio

NeurIPS 2021 · Architectures and efficiency › RNNs and normalization · approach: Mechanistic circuit interpretability · 12 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 discretized communication (DVNC) to transformer attention outputs, G=8Transformer · Sort-of-CLEVR binaryaccuracy: 76.00 -> 88.00Improved
add DVNC to transformer attention outputs, G=8Transformer · Sort-of-CLEVR ternaryaccuracy: 57.25 -> 62.67Improved
add DVNC to transformer attention outputs, G=8Transformer · Sort-of-CLEVR unaryaccuracy: 97.75 -> 98.75No clear effect
use a single discretization head (G=1) instead of G=8 (ablation)DVNC transformer · Sort-of-CLEVR binaryaccuracy: 88.00 -> 80.50Hurt
increase heads from G=8 to G=16 (ablation)DVNC transformer · Sort-of-CLEVR binaryaccuracy: 88.00 -> 84.00Hurt
add DVNC to GNN edge communicationGNN · Atari OOD movement prediction (Alien, Pong, SpaceInvaders)score: 0.1991 -> 0.2876 (Alien)Improved
add DVNC to GNN edge communicationGNN · Atari OOD Boxingscore: 0.9228 -> 0.9502No clear effect
add DVNC to GNN edge communicationGNN · Atari OOD Berzerkscore: 0.6077 -> 0.6233No clear effect
discretize communication results rather than edge hidden states (ablation)GNN with DVNC · 2D/3D shapes OODHITsImproved
discretize attention communication results rather than recurrent updates or inputs (ablation)RIMs with DVNC · Adding task OODOOD performanceImproved
use VQ-VAE nearest-neighbor discretization instead of Gumbel-Softmax (ablation)RIMs DVNC · Adding task OODOOD performanceImproved

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