AI Paper Atlas

Research map / Large language models / Retrieval and knowledge

MoL: Adaptive Mixture-of-Length Reasoning for Efficient Question Answering with Context

Guocong Li, Jinjian Zhang, Ping Wang, Dongnan Liu, Tian Liang, Qiuyi Qi, Hao Huang, Siyan Guo, Mutian Bao, Wei Zhou, Linjian Mo, Hongxia Xu and 1 more

ICLR 2026 · Large language models › Retrieval and knowledge · approach: Explanatory and multiple-choice QA benchmarks

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 extension reward Rextend (ablation)MoL · Loongaccuracy: 58.9 -> 62.3Improved
add compression reward Rcompress (ablation)MoL · Loongoutput tokens: 2862 -> 1374Improved
use question-relevant key sentence difficulty instead of original difficulty labels (ablation)MoL · HotpotQAaccuracy: 63.0 -> 67.2Improved
use key-sentence difficulty instead of whole-passage similarity (ablation)MoL · HotpotQAoutput tokens: 594 -> 316Improved
add progressive learning strategy lambda(t) (ablation)MoL · Loongaccuracy: 49.1 -> 62.3Improved
add progressive learning strategy lambda(t) (ablation)MoL · HotpotQAoutput tokens: 835 -> 316Improved
raise alpha from 0.1 to 0.9MoL · HotpotQAaccuracy and tokens: 64.9 -> 67.3 accuracy, 264 -> 371 tokensMixed
raise gamma from 0.3 to 0.9MoL · HotpotQAaccuracy: 67.2 -> 65.1Hurt
raise epsilon1 from 0 to 0.8MoL · HotpotQAaccuracy: 67.4 -> 65.1Hurt
set epsilon2 to 0.6 rather than 0MoL · HotpotQAaccuracy: 61.7 -> 67.2Improved
vary initial Ltarget (256 to 1024 sets)MoL · HotpotQAaccuracy: 66.7 to 67.4No clear effect
apply MoL instead of GRPO on Qwen3-8BQwen3-8B · HotpotQAaccuracy and tokens: 63.7 -> 67.2 accuracy, 747 -> 316 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