AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Don’t Start Over: A Cost-Effective Framework for Migrating Personalized Prompts Between LLMs

Ziyi Zhao, Chongming Gao, Yang Zhang, Haoyan Liu, Weinan Gan, Huifeng Guo, Yong Liu, Fuli Feng

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 7 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
migrate soft prompts with PUMA adapter instead of retraining from scratchpersonalized soft prompts · Amazon (1B->3B migration)RMSE: 0.9414 -> 0.9135Improved
migrate soft prompts with PUMA adapter instead of retrainingpersonalized soft prompts · MIND (1B->3B migration)uAUC: 0.5289 -> 0.6552Improved
migrate with PUMA instead of retrainingpersonalized soft prompts · Amazontraining time (h): 24.0 -> 0.48Improved
use K-means + variance stratification instead of random sampling at same budget (ablation)PUMA user selection · Amazon (2k-user selection)RMSE: 0.9419 -> 0.9315Improved
use K-means + variance stratification instead of K-means stratified alone (ablation)PUMA user selection · Amazon (2k-user selection)RMSE: 0.9546 -> 0.9315Improved
use K-means + variance stratification instead of random on Yelp (ablation)PUMA user selection · Yelp (2k-user selection)RMSE: 1.1146 -> 1.1111No clear effect
use K-means + variance stratification vs Loss Bucketing (ablation)PUMA user selection · MIND (1.5k-user selection)uAUC: 0.5845 -> 0.6344Improved
aggregate prompts from two source models (Llama + StableLM)PUMA migration to Phi-3 · Amazon aggregated migration to Phi-3RMSE: 0.9293 -> 0.9217Improved
chain migration across five modelsPUMA · Amazon chain migration (Llama start)RMSE: 0.9348 -> 0.9277Improved
migrate from weaker source model to stronger targetPUMA · Amazon cross-architecture (Figure 4)RMSE gainMixed

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