AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

xLSTM Scaling Laws: Competitive Performance with Linear Time-Complexity

Maximilian Beck, Kajetan Schweighofer, Sebastian Böck, Sebastian Lehner, Sepp Hochreiter

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Neural scaling laws · 7 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
use xLSTM instead of TransformerLLM pretraining · validation loss vs FLOPs, ~5 orders of magnitude of computecross-entropy lossImproved
use xLSTM instead of TransformerLLM inference · 16k prefilltime to first token: 30-50% lower TTFT for the same model sizeImproved
use xLSTM instead of TransformerLLM inference · 16k prefillstep timeImproved
increase token-to-parameter ratio into over-training (ablation)xLSTM · loss vs compute power lawexponent eta: 0.047 (M=22) -> 0.047 (M=1100)No clear effect
increase token-to-parameter ratio into over-training (ablation)Transformer · loss vs compute power lawexponent eta: 0.050 (M=22) -> 0.049 (M=550)No clear effect
use xLSTM instead of Transformercompute-optimal model size · IsoFLOPoptimal parameters N*Improved
increase context length from 2048 to 16384 (ablation)Transformer · IsoFLOPcompute-optimal model sizeHurt
increase context length from 2048 to 16384 (ablation)xLSTM · IsoFLOPcompute-optimal model sizeNo clear effect
include gamma as a free parameter in the loss surface fit (ablation)parametric loss fit · Token/Param runsfit qualityImproved

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