Metraipolis › archive › dc_0bf2a860640d

CPR for LLMs: Critical-Point Routing against Catastrophic Forgetting in Domain Adaptation — arXiv:2608.30158

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.30158
License (as archived)
CC0-1.0
Archived
2026-10-17T07:52:27.805Z
SHA-256 of the archived text
b99c89977d9c0cb880df646e1b2740812f59f6e92acbb33c3d753bde90276244
Stable URL of this page
https://metraipolis.com/documento/dc_0bf2a860640d

How to cite

CPR for LLMs: Critical-Point Routing against Catastrophic Forgetting in Domain Adaptation — arXiv:2608.30158 — arXiv (en), https://arxiv.org/abs/2608.30158 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_0bf2a860640d (Metraipolis archive dc_0bf2a860640d, sha256 b99c89977d9c0cb880df646e1b2740812f59f6e92acbb33c3d753bde90276244)
BibTeX
@misc{dc0bf2a860640d,
  title = {CPR for LLMs: Critical-Point Routing against Catastrophic Forgetting in Domain Adaptation — arXiv:2608.30158},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.30158},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_0bf2a860640d; sha256 b99c89977d9c0cb880df646e1b2740812f59f6e92acbb33c3d753bde90276244},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30158. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Supervised fine-tuning (SFT) is the de facto standard for adapting large language models (LLMs) to target domains, but it often degrades the model's general capabilities, a phenomenon known as catastrophic forgetting. Existing approaches typically modify the SFT loss to mitigate forgetting, but they inevitably operate along a domain-generality trade-off. In this work, we step outside this trade-off by decoupling the two capabilities at the model level: we keep the original base model for general capability, and selectively invoke the SFT expert only when domain-specific knowledge is required. Specifically, we propose CPR (Critical-Point Routing), a token-level routing framework between a base model and its expert derivative, based on critical tokens where the base model fails but the expert succeeds. We train a lightweight hierarchical router that estimates the expert-call probability per token, and pair it with a tailored inference procedure that combines momentum smoothing and threshold gating. Across diverse model-domain configurations, CPR achieves state-of-the-art across all settings, surpassing SFT expert by 1.4-5.5% in domain performance while recovering its general-capability drop from 3.4-14.5% to at most 0.5%, with minimal overhead from invoking the expert on only one-third of tokens.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2608.30158 — 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 2026-10-17T07:28:26.833Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).