AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Exploring Depth Generalization in Large Language Models for Solving Recursive Logic Tasks

Zhiyuan He

AAAI 2026 · Architectures and efficiency › Attention and transformers · approach: Attention head mechanisms · 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 looped locate-and-replace instead of end-to-endGPT-2 · Boolean algebra, depth 12accuracy: 51.8% -> 66.7%Improved
use looped locate-and-replace instead of end-to-endGPT-2 · Compositional arithmetic and propositional logicaccuracyImproved
remove locate-and-replace from looped pipeline (Looped Only) (ablation)LLR · Boolean algebraaccuracyImproved
increase model size from 4 layers to 8 layers (ablation)GPT-2 · Boolean algebra depth up to 10accuracy: difference never exceeds 1%No clear effect
use ALiBi instead of NoPE/absolute/RoPE for locator (ablation)LLR locator · length 20 generalizationaccuracy: ALiBi 96%Improved
increase recursion depth vs increase lengthEnd-to-end GPT-2 · Boolean algebraaccuracy: PCC depth -0.9210 vs length -0.2068Hurt

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