AI Paper Atlas

Research map / Adversarial robustness and security / Attacks on text and graphs

CoSPED: Consistent Soft Prompt Targeted Data Extraction and Defense

Zhuochen Yang, Kar Wai Fok, Vrizlynn L. L. Thing

AAAI 2026 · Adversarial robustness and security › Attacks on text and graphs · approach: LLM jailbreak and injection defense · 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 Additive loss to MLE and Smooth loss (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: +1.3%Improved
add Dynamic loss to MLE loss (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: +0.8%Improved
use Focal loss with Smooth loss instead of MLE (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: +0.6%Improved
combine multiple extra losses rather than a single loss (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: 61.2 -> 64.9Improved
Self Consistency decoding instead of beam search (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: 64.5 -> 64.2No clear effect
Self Consistency decoding instead of greedy decoding (ablation)CoSPED soft prompt · LM Extraction Benchmark D1ER50: 58.7 -> 64.2Improved
increase prefix length from 50 to 100 tokens (20-token soft prompt) (ablation)CoSPED · LM Extraction Benchmark D2ER50: +23.8%Improved
increase soft prompt length from 20 to 100 tokens (50-token prefix) (ablation)CoSPED · LM Extraction Benchmark D2ER50: +8.8%Improved
increase soft prompt length beyond 100 tokens (ablation)CoSPED · LM Extraction Benchmark D2ER50: 64.5 -> 62.1 (50-token prefix, 100 -> 150 tokens)No clear effect
use CoSPED over Ethicist across soft prompt and prefix lengthsGPT-Neo 1.3B · LM Extraction Benchmark D2ER50: 62.3 -> 65.2 (D1)Improved
apply modified ROME editing as a defenseGPT-Neo 1.3B under CoSPED attack · 1000 test prefix-suffix pairsER50: 63.9 -> 0Improved
stop ROME editing at training PPL 10 instead of PPL 6 (ablation)ROME defense on GPT-Neo 1.3B · LAMBADALB accuracy: 0.2598 -> 0.1247Hurt

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