AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

COBRA: Contribution-Based Bayesian Rank Allocation for Parameter-Efficient Fine-Tuning

Hongcheng Ding, Xuanze Zhao, Xuanhuang Liu, Jing Jin, Shamsul Nahar Abdullah, Deshinta Arrova Dewi

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
replace uniform rank (LoRA r=8) with COBRA heterogeneous ranksLoRA · GLUE avg, RoBERTa-baseAvg. score: 87.0 -> 88.6Improved
use COBRA instead of AdaLoRAAdaLoRA · GLUE avg, DeBERTaV3-baseAvg. score: 89.0 -> 89.9Improved
use COBRA instead of gradient-driven GoRAGoRA · T5-Base GLUE-5Avg. score: 88.0 -> 89.1Improved
use COBRA instead of GoRAGoRA · LLaMA-3.1-8B HumanEvalAccuracy: 49.0 -> 52.7Improved
use COBRA instead of uniform LoRA in high-rank regimeLoRA r=384 · NEU financial sentimentMSE: 0.0189 -> 0.0177Improved
use multiplicative (Phi x Gamma) instead of additive aggregation (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 89.2 -> 89.9Improved
combine contribution and adaptation demand instead of a single factor (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 88.8 -> 89.9Improved
use informative TA-LC prior (gamma=5) instead of uniform prior (gamma=0) (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 88.5 -> 89.9Improved
increase prior concentration to gamma=50 (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 89.9 -> 88.3Hurt
add KL regularization (beta=0.01) vs none (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 88.8 -> 89.9Improved
use strong KL weight beta=1.0 (ablation)COBRA · GLUE avg, DeBERTaV3-baseAvg. score: 89.9 -> 88.2Hurt
use DyLoRA range-based dynamic allocation instead of uniform LoRALoRA · DeBERTaV3-base GLUE r=4Avg. score: 88.7 -> 88.4Hurt

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