AI Paper Atlas

Research map / Architectures and efficiency / Pruning and efficient tuning

Put the Space of LoRA Initialization to the Extreme to Preserve Pre-trained Knowledge

Pengwei Tang, Xiaolin Hu, Yong Liu, Lizhong Ding, Dongjie Zhang, Xing Wu, Debing Zhang

AAAI 2026 · Architectures and efficiency › Pruning and efficient tuning · approach: LoRA variants and theory · 20 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
initialize LoRA in the null space of activations instead of CorDA initializationLoRA · Math, LLaMA-2-7BAvg1(Per): 76.24 -> 79.21Improved
initialize LoRA in the null space of activations instead of CorDA initializationLoRA · Code, LLaMA-2-7BAvg1(Per): 91.42 -> 95.12Improved
initialize LoRA in the null space of activations instead of CorDA initializationLoRA · Instruction following, LLaMA-2-7BAvg1(Per): 87.36 -> 89.90Improved
initialize LoRA in the null space of activations instead of CorDA initializationLoRA · Math, LLaMA-3.2-3BAvg1(Per): 89.00 -> 93.18Improved
initialize LoRA in the null space of activations instead of CorDA initializationLoRA · Math, LLaMA-3.2-3BAvg2 (downstream): 36.81 -> 36.41Hurt
use LoRA-Null initialization instead of PiSSALoRA · Math, LLaMA-2-7BAvg2 (downstream): 29.67 -> 26.62Hurt
reduce calibration set to 64 samples (CorDA) (ablation)CorDA · Math, LLaMA-3.2-3BAvg1(Per): 89.00 -> 68.97Hurt
reduce calibration set to 64 samples (LoRA-Null) (ablation)LoRA-Null · Math, LLaMA-3.2-3BAvg1(Per): 93.00 -> 94.95No clear effect
increase LoRA rank from 128 to 256 (CorDA) (ablation)CorDA · Math, LLaMA-3.2-3BAvg1(Per): 89.00 -> 73.01Hurt
increase LoRA rank from 128 to 256 (LoRA-Null) (ablation)LoRA-Null · Math, LLaMA-3.2-3BAvg1(Per): 93.00 -> 84.71Hurt
scale singular values of LoRA-Null init by 4 (ablation)LoRA-Null · Math, Gemma-3-1BAvg1(Per): 80.78 -> 65.98Hurt
scale singular values of LoRA-Null init by 4 (ablation)LoRA-Null · Math, Gemma-3-1BGSM8k: 16.91 -> 22.90Improved

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