AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Stacked from One: Multi-Scale Self-Injection for Context Window Extension

Wei Han, Pan Zhou, Shuicheng YAN

ICLR 2026 · Large language models › Language, safety and interpretability · approach: LLM pruning and scaling data · 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
use the default right-branch policy instead of random branch selection (ablation)SharedLLM · Arxiv-32K perplexityperplexity: 2.61 -> 2.46Improved
use default policy instead of fixed-left selection (ablation)SharedLLM · Arxiv-32K perplexityperplexity: 2.49 -> 2.46Improved
use default policy instead of query-aware selection for language modeling (ablation)SharedLLM · Arxiv-32K perplexityperplexity: 2.48 -> 2.46Improved
use default policy instead of Lambda-shape policy (ablation)SharedLLM · Arxiv-32K perplexityperplexity: 2.52 -> 2.46Improved
use default policy instead of random selection on multi-document QA (ablation)SharedLLM · MD-QA (LLaMA-2 SFT)F1: 28.85 -> 30.93Improved
add query-aware information gathering, noise in node splitting, chunk-level positional indices (ablation)SharedLLM · Design-choice ablationperformanceImproved
inject context into the bottom M layers rather than other layer choices (ablation)SharedLLM · Injection layer ablationperformanceImproved
train with Books3 (data) vs without (ablation)CEPE · Arxiv 4K perplexityperplexity: 3.03 -> 2.86Improved
replace CEPE with SharedLLMCEPE · Arxiv 4K perplexityperplexity: 3.03 -> 2.99Improved
replace Activation Beacon with SharedLLMActivation Beacon · MD-QA (LLaMA-2 SFT)F1: 28.44 -> 30.93Improved
replace Activation Beacon with SharedLLMActivation Beacon · InfBench Ret.N (LLaMA-2 SFT)score: 80.58 -> 82.79Improved
add query-aware retrieval (ablation)SharedLLM · Inference time (8K tokens)time (s) / memory (GB): 11.81 -> 13.18 s; 22.39 -> 25.44 GBHurt

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