Metraipolis › archive › dc_ecf84e692fda

KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise — arXiv:2610.41480

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.41480
License (as archived): CC0-1.0
Archived: 2026-11-03T07:45:27.601Z
SHA-256 of the archived text: 1350947281a6b6f0f60b7f6bee6c72808e6410e278917290084271b72a4d79b0
Stable URL of this page: https://metraipolis.com/documento/dc_ecf84e692fda

How to cite

KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise — arXiv:2610.41480 — arXiv (en), https://arxiv.org/abs/2610.41480 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ecf84e692fda (Metraipolis archive dc_ecf84e692fda, sha256 1350947281a6b6f0f60b7f6bee6c72808e6410e278917290084271b72a4d79b0)

BibTeX

@misc{dcecf84e692fda,
  title = {KV-Cache Patching: Tracing How and Where Fine-Tuned Behaviors Arise — arXiv:2610.41480},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.41480},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ecf84e692fda; sha256 1350947281a6b6f0f60b7f6bee6c72808e6410e278917290084271b72a4d79b0},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.41480. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Fine-tuning a language model changes how it responds, but where is this fine-tuned behavior instantiated? To better understand this, we evaluate whether the change in model behavior arises only during the generation of the answer tokens where the change manifests, or whether altered processing of previous tokens causes it via the attention mechanism. For this, we patch the KV cache from a LoRA fine-tuned model into the unmodified base model, and vice versa. Our results show that for some behaviors, like responding in German or emergent misalignment, patching the KV cache can indeed transfer them from the fine-tuned to the base model. Correspondingly, patching the KV cache from base to fine-tune can partially or fully remove fine-tuned behavior. Furthermore, we find that the effect of patching grows with how much of the KV cache is patched and concentrates at a few pivotal positions, with system-prompt and chat-template tokens having disproportionate influence.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2610.41480 — 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 2026-11-03T07:39:52.172Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).