Crimson AI NewsA CrimsonLingua Network service
EN ع
← Back to news
Research paper Hugging Face

AdaPop: Adaptive Popularity Boosts LLM Unlearning, Cutting Leakage by 5x

AI By Crimson AI Hugging Face Papers 20 August 2026 · 00:00 8 views
Share: X Telegram

Hugging Face researchers propose AdaPop, an unlearning method that adapts gradient pressure based on fact popularity, reducing leakage of forgotten content by up to 5x under paraphrased queries.

AdaPop: Adaptive Popularity Boosts LLM Unlearning, Cutting Leakage by 5x

Key points

Researchers at Hugging Face have introduced AdaPop (Adaptive Popularity), a novel method for machine unlearning in large language models (LLMs). The approach addresses a key limitation of existing techniques: they apply uniform gradient pressure regardless of how frequently a fact appeared in training data. However, popular facts are memorized more deeply and resist removal longer than rare ones.

AdaPop combines local token confidence with a per-fact popularity-dependent exponent, derived from external proxies such as Wikidata sitelinks or an LLM-as-Judge. It also automates the delicate balance between forgetting and retaining knowledge using a dual-ascent controller that adjusts the retain penalty each epoch.

In evaluations across three model families and two benchmarks, AdaPop leaked approximately 5 times less forgotten content than competing methods under paraphrased queries, and about 1.6 times less under adversarial reformulations. Internal metrics further showed that forget-set hidden states moved further away from the pre-unlearning model's states, while retain-set representations remained close.

This work highlights the importance of considering data frequency in unlearning, offering a more effective approach to privacy and compliance in AI systems.

ScenarioLeakage Reduction (AdaPop vs. baselines)
Paraphrased queries~5x less
Adversarial reformulations~1.6x less
Source
Hugging Face · Hugging Face Papers
Related news
Research paper
Hugging Face 29 Aug 2026

Hugging Face Audit: 110 of 124 AI Evaluations Fail to Support Their Claims

A new commit-bound census of 124 Inspect Evals units reveals that 110 stop before deterministic inference due to missing historica...

4
Research paper
Hugging Face 29 Aug 2026

Aphanta: New Framework Diagnoses When Image Editing Boosts Multimodal Reasoning

Hugging Face researchers introduce Aphanta, a diagnostic framework that evaluates when image-editing intermediates improve multimo...

5
Research paper
Hugging Face 29 Aug 2026

EditaLive! Enables Real-Time Character Video Editing for Live Streaming

Hugging Face researchers introduce EditaLive, a framework for real-time human-centric video editing in live streams, achieving sta...

4