AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

ParaDySe: A Parallel Strategy Switching Framework for Dynamic Sequences in Transformer-based Large Language Models

Zhixin Ou, Peng Liang, Linbo Qiao, Jianchen Han, Baihui Liu

AAAI 2026 · Architectures and efficiency › Attention and transformers · approach: Sparse attention for long-context LLMs · 1 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 ParaDySe adaptive strategy switching instead of static baselines on GitHubCodeBERT training · GitHubCode max sequencestraining time: up to 89.29% less timeImproved
use ParaDySe instead of baselines on GitHubCodeGPT (largest) · GitHubCodemax trainable sequence length: up to +144.16%Improved
use ParaDySe instead of baselines on GRCh38BERT / GPT · GRCh38training time / sequence length: 58.01% faster (BERT); 181.22% longer (GPT)Improved
use ParaDySe instead of baselines for LLaMA on one caseLLaMA · long-sequence datasetsmax sequence lengthMixed
add MegatronTS to the strategy set (ablation)ParaDySe · GPT on GRCh38time saving: 300032 -> 329728 tokens; 11.80% time savingImproved
add METP to the strategy set (ablation)ParaDySe · GPT on GRCh38max sequence length: 112128 -> 329728 tokensImproved
add MegatronCZ to the strategy set (ablation)ParaDySe · GPT on GRCh38time saving: -0.13% saving; seq len 300032 -> 329728No clear effect
add random forest to the polynomial cost model (hybrid) (ablation)ParaDySe cost model · GPT on GRCh38max sequence length: 270336 -> 329728 tokens; 1.29% savingImproved
add smoothing of strategy switching (ablation)ParaDySe · GPT on GRCh38time saving: 0.04% saving; 300032 -> 329728 tokensImproved

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