AI Paper Atlas

Research map / Large language models / Instruction and fine-tuning

BA-LoRA: Bias-Alleviating Low-Rank Adaptation to Mitigate Catastrophic Inheritance in Large Language Models

Yupeng Chang, Yi Chang, Yuan Wu

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Low-rank adaptation variants · 14 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
add consistency regularizer (ablation)PiSSA · GSM8Kaccuracy: 51.48 -> 54.25Improved
add diversity regularizer (ablation)PiSSA · GSM8Kaccuracy: 51.48 -> 53.60Improved
add SVD regularizer (ablation)PiSSA · GSM8Kaccuracy: 51.48 -> 52.95Improved
add all three regularizers (ablation)PiSSA · GSM8Kaccuracy: 51.48 -> 55.86Improved
add consistency regularizer (ablation)PiSSA · MATHaccuracy: 7.60 -> 9.15Improved
add all three regularizers (ablation)PiSSA · GLUE average (DeBERTa-v3-base)average score: 89.47 -> 90.67Improved
add three regularizers (ablation)LoRA · GSM8Kaccuracy: 42.68 -> 51.82Improved
add three regularizers (ablation)DoRA · GSM8Kaccuracy: 41.77 -> 52.71Improved
add three regularizers (ablation)LoRA · MBPPPass@1: 21.51 -> 33.81Improved
use covariance diversity regularizer instead of entropy (ablation)BA-LoRA · MNLIaccuracy: 90.41 -> 91.26Improved
BA-LoRA vs PiSSA on noisier-pretrained T5-baseBA-LoRA · GLUE subsetaverage score: 84.71 -> 87.97Improved
BA-LoRA vs PiSSA on RoBERTa-baseBA-LoRA · GLUE subsetaverage score: 85.23 -> 86.34Improved

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