AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

KDP: Simplifying Representation Dynamics in Kernel Space

Zeyu Ma, Wanying Wang, Guchu Zou, Mingang Chen, Jianhong Wu

ICLR 2026 · Architectures and efficiency › RNNs and normalization · approach: Layer-wise analysis of language models

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
fold blocks via kernel-space linearization instead of simply removing layers (ablation)KDP on LLaMA-2-7B · 12 classification benchmarks avgaverage accuracy: 34.28 -> 53.11Improved
fold blocks via kernel-space linearization instead of simply removing layers (ablation)KDP on LLaMA-2-13B · 12 classification benchmarks avgaverage accuracy: 45.01 -> 59.89Improved
fold blocks via kernel-space linearization instead of simply removing layers (ablation)KDP on LLaMA-3-8B · 12 classification benchmarks avgaverage accuracy: 39.58 -> 52.77Improved
linearize in learnable RFF kernel space rather than original space with OLS (ablation)KDP on LLaMA-3-8B · HellaSwagaccuracy: 31.3 -> 61.5Improved
linearize in learnable RFF kernel space rather than original space with OLS (ablation)KDP on LLaMA-3-8B · MMLUaccuracy: 26.7 -> 55.5Improved
replace more consecutive layers in one block (larger Kmax) (ablation)KDP on LLaMA-2-7B · MMLUaccuracy: 48.5 -> 24.0Hurt
increase RFF kernel size from 256 to 1024 (ablation)KDP on LLaMA-2-7B · HellaSwagaccuracy: 33.7 -> 65.1Improved
add PEFT (output-head LoRA with distillation) after KDPKDP on LLaMA-3.1-8B · WikiTextperplexity: 205.7 -> 31.7Improved
add PEFT after KDPKDP on LLaMA-2-7B · 12 classification benchmarks avgaverage accuracy: 53.11 -> 52.52No clear effect
increase pruning ratio (ablation)KDP · CMMLUretained performanceHurt

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