AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

IGU-LoRA: Adaptive Rank Allocation via Integrated Gradients and Uncertainty-Aware Scoring

Xuan Cui, HuiYue Li, Run Zeng, Yunfei Zhao, Jinrui Qian, Wei Duan, Bo Liu, Zhanpeng Zhou

ICLR 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory · 5 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
replace instantaneous-gradient importance with parameter-space integrated gradients plus uncertainty-aware (EMA SNR) scoringadaptive-rank LoRA (vs AdaLoRA) · GLUE average (RoBERTa-large)average score: 88.68 -> 89.42Improved
IGU-LoRA vs AdaLoRA on reasoning suiteQwen-2.5-0.5B · 5-task reasoning averageaverage accuracy: 57.67 -> 59.17Improved
IGU-LoRA vs full fine-tuningQwen-2.5-0.5B · GSM8Kaccuracy: 34.64 -> 34.16Hurt
remove the sampled alpha integration coefficient (ablation)IGU-LoRA · BoolQ+GSM8K avgaverage accuracy: 57.82 -> 58.31Improved
use N=20 alpha candidates instead of N=4 (ablation)IGU-LoRA · BoolQ+GSM8K avgaverage accuracy: 57.93 -> 58.31Improved
use SNR (EMA mean / deviation) instead of AdaLoRA-style multiplicative sensitivity-uncertainty (ablation)IGU-LoRA · BoolQ+GSM8K avgaverage accuracy: 57.99 -> 58.31Improved
IGU-LoRA vs DoRA/LoRA on Llama-2-7BLlama-2-7B · 5-task averageaverage accuracy: 89.91 (AdaLoRA) -> 90.44Improved
IGU-LoRA vs LoRA on DeepSeek-R1-Distill-Qwen-2.5-7B GSM8KDeepSeek-R1-Distill-Qwen-2.5-7B · GSM8Kaccuracy: 74.60 -> 74.28No clear effect
remove high-score modules (L3_K and L10_V) vs low-score modules (ablation)IGU-LoRA rank allocation · BoolQaccuracy: 82.45 -> 80.44 (high-rank removed); 82.45 -> 82.30 (low-rank removed)Hurt
add integrated gradients and SNR scoring (extra overhead vs LoRA)LoRA · Qwen-2.5-0.5B BoolQ trainingtraining time (h): 0.42 -> 0.87Hurt
IGU-LoRA vs LoRA on MMLULlama-2-7B · MMLU averageaverage accuracy: 49.94 -> 51.07Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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