AI Paper Atlas

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

Not All Tokens Are Meant to Be Forgotten

Xiangyu Zhou, Yao Qiang, Saleh Zare Zade, Douglas Zytko, Prashant Khanduri, Dongxiao Zhu

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 5 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 unwanted-information identifier (GPT) to baselines GA, NPO, KTO, SimNPO (ablation)GA/NPO/KTO/SimNPO · TOFU (1%, 5%, 10% forget)model utilityImproved
add unwanted-information identifier to NPO (NPO-GPT) alone (ablation)NPO · TOFU forget05model utility: 74% utility decline remainsMixed
add Preservation Loss on general words to NPO-GPT (ablation)NPO-GPT · TOFU forget05model utilityImproved
replace NPO loss with Logit Preference Loss (TPO) (ablation)NPO+PL-GPT · TOFU forget05model utility and KL divergenceImproved
use NPO with the entire forget sample as target (larger forget set)NPO · TOFUmodel utility: 0.76 -> 0.11Hurt
increase forget set size to 10%GA and KTO based methods · TOFU forget 10%forget quality: near-zero forget qualityHurt
use TPO-GPT at 10% forget setTOFU LLaMA-2 7B and LLaMA-3.2 3B · TOFU forget 10%forget quality and utility: 70% utility with highest forget qualityImproved
add GDR retain loss to unlearning methods (ablation)GA/NPO/SimNPO/TPO · MUSE News and Books (LLaMA-2 7B)KnowMem on Dr: TPO 0.00 -> 43.67 (News), 0.00 -> 42.07 (Books)Improved
use TPOGDR instead of SimNPOGDRMUSE News · MUSE News (LLaMA-2 7B)PrivLeak: -97.17 -> -6.12Improved
use Task Vector baselineTask Vector · MUSEforgetting: VerbMem 66.74 vs original 56.26 (News)Hurt
use ChatGPT-4 generative identifier instead of DistilBERT discriminative identifier (ablation)TIF · TOFUforget quality vs utility balanceImproved

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