AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

It's All Connected: A Journey Through Test-Time Memorization, Attentional Bias, Retention, and Online Optimization

Ali Behrouz, Meisam Razaviyayn, Peilin Zhong, Vahab Mirrokni

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Brain-inspired recurrent architectures · 66 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 retention gate (ablation)Yaad · Average LM score (Yaad ablation)avg LM score: 50.63 -> 53.98Improved
make Huber threshold delta input-dependent (ablation)Yaad · Average LM score (Yaad ablation)avg LM score: 52.19 -> 53.98Improved
use Huber loss instead of pure l2 (ablation)Yaad · Average LM score (Yaad ablation)avg LM score: 52.86 -> 53.98Improved
use Huber loss instead of pure l1 (ablation)Yaad · Average LM score (Yaad ablation)avg LM score: 53.04 -> 53.98Improved
use 2-layer MLP memory instead of linear memory (ablation)Yaad · Average LM score (Yaad ablation)avg LM score: 51.57 -> 53.98Improved
replace Gated DeltaNet with Moneta1.3B model · WikiTextperplexity: 16.42 -> 15.52Improved
replace Gated DeltaNet with Yaad1.3B model · WikiTextperplexity: 16.42 -> 15.18Improved
use Moneta instead of Gated DeltaNetS-NIAH · S-NIAH averageaccuracy: 75.8 -> 93.5Improved
use Memora instead of Gated DeltaNet760M model · WikiTextperplexity: 21.18 -> 22.28Hurt
vary p of lp attentional bias (p=4 vs p=3) (ablation)Moneta · LM across 2K-16K contextperformanceMixed

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