AI Paper Atlas

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

From Prompts to Responses: Dual-Sided Data Leakage and Defense in Split Large Language Models

Zixuan GU, Xiaojun Ye, Yang Liu

ICML 2026 · Large language models › Instruction and fine-tuning · approach: Calibration and knowledge in LLM tuning · 0 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 Dual-Sided Initialization (DSI) in place of vanilla init (ablation)PIDI on Llama3-8B · FinAP(alpha=0.5): 0.697 -> 0.883Improved
add Patched Model Inversion (PMI) on top of DSI (ablation)DSI+VMI · Fin Llama3-8BAP(alpha=0.5): 0.716 -> 0.883Improved
add PMI to DSI on Qwen2.5-7B (ablation)DSI+VMI · DollyAP(alpha=0.5): 0.814 -> 0.985Improved
add PMI to DSI on Llama3.2-3B Med (ablation)DSI+VMI · MedAP(alpha=0.5): 0.925 -> 0.901Hurt
use DSI initialization instead of MC initialization with VMI (ablation)VMI · Fin Llama3-8BAP(alpha=0.5): 0.423 -> 0.716Improved
increase patch length l (ablation)PMI · Fin Llama3.2-3BAP_inputHurt
deepen split point from 3-3 to 6-6 without defense (ablation)PIDI attack · Fin Llama3.2-3BAP(alpha=0.5): 0.950 -> 0.726Hurt
reduce attacker auxiliary dataset size (ablation)PIDI · FinAPNo clear effect
apply ADMI defensePIDI attack on Llama3.2-3B · FinAP(alpha=0.5): 0.868 -> 0.017Improved
apply ADMI defenseLlama3.2-3B · FinMP (METEOR): 0.56 -> 0.516Hurt
use ADMI instead of DualGuardPIDI defense on Qwen2.5-7B · FinAP(alpha=0.5): 0.572 -> 0.004Improved
apply ADMI defenseQwen2.5-7B · Dollyinference throughput (token/s/GPU): 24.892 -> 20.418Hurt

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