Metraipolis › archive › dc_ce70870b5deb
Preserving Utility in Continual LLM Unlearning via Orthogonal Projections — arXiv:2610.40483
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.40483
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-31T06:17:27.099Z
- SHA-256 of the archived text
527477dcc8e2bc5a638a043fae605c5a486f58e897801b14b695df9c881fd589- Stable URL of this page
https://metraipolis.com/documento/dc_ce70870b5deb
How to cite
Preserving Utility in Continual LLM Unlearning via Orthogonal Projections — arXiv:2610.40483 — arXiv (en), https://arxiv.org/abs/2610.40483 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ce70870b5deb (Metraipolis archive dc_ce70870b5deb, sha256 527477dcc8e2bc5a638a043fae605c5a486f58e897801b14b695df9c881fd589)
BibTeX
@misc{dcce70870b5deb,
title = {Preserving Utility in Continual LLM Unlearning via Orthogonal Projections — arXiv:2610.40483},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.40483},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ce70870b5deb; sha256 527477dcc8e2bc5a638a043fae605c5a486f58e897801b14b695df9c881fd589},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.40483. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Unlearning concerns the task of removing the influence of a sensitive portion of the training data, such as protected, copyrighted or private text, from a model without retraining it from scratch. While unlearning requests happen continuously during a model's lifetime, the vast majority of LLM unlearning methods are ill-suited for iterated removal, yielding steep decay in forgetting and utility upon sequential unlearning. This decay stems from interference between successive updates, as each new deletion can undo earlier ones or push the model further along directions already degraded by them. To address this, we introduce Continual Orthogonal Projection (COP), a gradient-projection scheme that restricts each update to the orthogonal complement of activation subspaces cached from prior forget sets. This approach provably guarantees first-order non-interference with earlier deletions. COP stores only a low-rank basis per layer, requiring neither past gradients nor previous forget sets, and plugs into unlearning methods that update the model through gradient steps on a differentiable loss without modifying its objective. Experiments on standard LLM unlearning benchmarks (TOFU and RWKU) show that COP preserves utility across sequential requests without weakening forgetting, preventing the collapse of standard objectives and forgetting more effectively than continual-specific baselines. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.40483 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-31T06:07:23.422Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).