AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Less Data, Faster Training: repeating smaller datasets speeds up learning via sampling biases

Jingwen Liu, Ezra Edelman, Surbhi Goel, Bingbin Liu

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Layer-wise analysis of language models · 0 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
train on a small random subset (2^14) instead of the full population (2^20) with full-batch GD2-layer MLP on (20,6) sparse parity · (20,6) sparse paritysteps and compute to converge: 1500 steps vs more than 2000 steps; 100x speedup in computeImproved
remove input bias (zero empirical mean of x, optionally y) from the small dataset (ablation)small-set training (2^14) on parity · (20,6) parityaccuracyNo clear effect
add input bias taken from a small set to online large-set training (ablation)online training · (20,6) parityaccuracy: speedup only large when fewer than 2^5=32 samplesMixed
use random labels in the first small-set phase before switching to large-set training (ablation)MLP · parity and SIMaccuracy / norm growthImproved
add per-layer init scaling alpha (first layer divided, second multiplied) (ablation)MLP full-population training · parity and SIMsmall-vs-large gapImproved
use muP parameterization (ablation)2-layer width-64 MLP · parity vs SIMsmall-vs-large gapMixed
add separate larger learning rate for W_Q, W_K (ablation)2-layer Transformer · (20,6) paritysmall-vs-large gap: optimal QK lr 3.6x (large set), 2.5x (100 epochs), 1.25x (6-phase)Improved
increase W_Q, W_K initialization scale (ablation)2-layer Transformer · (20,6) paritysmall-vs-large gap: optimal 8x default for large set; 2x for 100-epoch; no benefit for 6-phaseImproved
add QK normalization (ablation)Transformer · sparse parity and SIM vs ICL and mod additionoptimization speed / overfittingMixed

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