AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Adaptive Utilization of Low-Rank Adaptation via Conditioned Gating

Guang Yang, Changhao Guan, Chao Huang, Yufeng Chen, Kaiyu Huang

ICML 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory

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 token-wise utilization gating (ablation)U-LoRA · Math reasoning avgavg accuracy: 78.95 -> 83.97Improved
add sequence-level aggregation (ablation)U-LoRA · Math reasoning avgavg accuracy: 79.16 -> 83.97Improved
add bias-corrected EMA historical prior (ablation)U-LoRA · Math reasoning avgavg accuracy: 79.66 -> 83.97Improved
add sequence-level aggregation (ablation)U-LoRA · AQuAaccuracy: 31.89 -> 43.31Improved
use U-LoRA instead of increasing LoRA rank to 32LLaMA-3-8B · Math reasoning avgavg accuracy: 71.67 -> 76.02Improved
increase LoRA rank from 16 to 32 (ablation)LoRA · GLUE avg (RoBERTa-Base)avg score: 82.82 -> 83.60Improved
U-LoRA versus LoRA across ranks 2 to 32 (ablation)Qwen3-8B · Math reasoning avgavg accuracy: 77.60 -> 82.06 (r=2)Improved
apply U-LoRA to more target modules (Q to QKVOGUD) (ablation)Qwen3-8B · Math reasoning avgavg accuracy: 80.75 -> 84.20Improved
vary EMA momentum beta (0.9 to 0.9995) (ablation)U-LoRA · Math reasoning avgavg accuracy: 83.23 -> 84.17 (0.9 vs 0.995)No clear effect
U-LoRA instead of TopLoRA at 72B scaleQwen2.5-72B Base · Math reasoning avgavg accuracy: 83.97 -> 85.08Improved
U-LoRA instead of TopLoRA on instruct modelQwen2.5-14B-Instruct · Math reasoning avgavg accuracy: 79.58 -> 82.82Improved
U-LoRA adds inference latency relative to LoRA (non-mergeable gating)Qwen3-8B r=8 · Efficiency trade-offtask latency: 1.00x -> 1.39xHurt

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