AI Paper Atlas

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

PerFit: Exploring Personalization Shifts in Representation Space of LLMs

Jiahong Liu, Wenhao Yu, Quanyu Dai, Zhongyang Li, Jieming Zhu, Menglin Yang, Tat-Seng Chua, Irwin King

ICLR 2026 · Large language models › Instruction and fine-tuning · approach: Fine-tuning generalization and delta compression

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
use PerFit representation intervention instead of OPPU per-user LoRALlama2-7B personalization · LaMP-2Naccuracy: 0.810 -> 0.818Improved
use PerFit instead of OPPULlama2-7B personalization · LaMP-2Maccuracy: 0.600 -> 0.630Improved
use PerFit instead of OPPULlama2-7B personalization · LaMP-3MAE: 0.179 -> 0.179No clear effect
use PerFit instead of OPPULlama2-7B personalization · LaMP-7ROUGE-1: 0.539 -> 0.525Hurt
use PerFit instead of OPPULlama2-7B personalization · LaMP-4ROUGE-1: 0.191 -> 0.207Improved
use PerFit trainable parameters instead of OPPULlama2-7B personalization · LaMP tasksparameter reduction: 81.25% to 98.44%Improved
use PerFit instead of existing fine-tuning baselinesLlama2-7B personalization · training timetraining time: 17.0% to 35.8% reductionImproved
add Stage-2 personalized training to Stage-1 only (ablation)PerFit · LaMP-2Maccuracy: 0.466 -> 0.630Improved
add Stage-2 personalized training to Stage-1 only (ablation)PerFit · LaMP-2Naccuracy: 0.792 -> 0.818Improved
add Stage-1 collective training to Stage-2 only (P rank) (ablation)PerFit · LaMP-2Maccuracy: 0.599 -> 0.630Improved
raise Stage-2 only rank from P to C+P (ablation)PerFit · LaMP-2Maccuracy: 0.599 -> 0.620Improved
intervene at later single layers instead of earlier (ablation)PerFit · Movie Tagging and News Headlineaccuracy, ROUGEHurt

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