AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Dynamic Thinking-Token Selection for Efficient Reasoning in Large Reasoning Models

Zhenyuan Guo, Tong Chen, Wenlong Meng, Chen GONG, Xin Yu, Chengkun Wei, Wenzhi CHEN

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Sparse attention for long-context LLMs · 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
add local window tokens (ablation)DynTS (R1-Llama) · Six reasoning benchmarks (average)Pass@1: 59.0 -> 61.9Improved
add question tokens (ablation)DynTS (R1-Llama) · Six reasoning benchmarks (average)Pass@1: 39.8 -> 61.9Improved
add critical thinking tokens (ablation)DynTS (R1-Llama) · Six reasoning benchmarks (average)Pass@1: 59.6 -> 61.9Improved
add local window tokens (ablation)DynTS (R1-Qwen) · Six reasoning benchmarks (average)Pass@1: 61.6 -> 65.3Improved
add question tokens (ablation)DynTS (R1-Qwen) · Six reasoning benchmarks (average)Pass@1: 62.6 -> 65.3Improved
add critical thinking tokens (ablation)DynTS (R1-Qwen) · Six reasoning benchmarks (average)Pass@1: 63.5 -> 65.3Improved
select thinking tokens by predicted importance instead of randomly (ablation)DynTS (R1-Llama) · Six reasoning benchmarks (average)Pass@1: 42.2 -> 61.9Improved
select thinking tokens by predicted importance instead of bottom-k (ablation)DynTS (R1-Qwen) · Six reasoning benchmarks (average)Pass@1: 57.6 -> 65.3Improved
select thinking tokens by predicted importance instead of randomly (ablation)DynTS (R1-Qwen) · Six reasoning benchmarks (average)Pass@1: 61.9 -> 65.3Improved
retaining random or bottom thinking tokens instead of removing all thinking tokens (ablation)DynTS (R1-Llama) · Six reasoning benchmarks (average)Pass@1: 59.6 -> 39.8Hurt
replace R-KV with DynTSR1-Llama KV compression · Six reasoning benchmarks (average)Pass@1: 59.6 -> 61.9Improved
replace SnapKV with DynTSR1-Qwen KV compression · Six reasoning benchmarks (average)Pass@1: 63.2 -> 65.3Improved

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