Metraipolis › archive › dc_debdee463029

Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs — arXiv:2609.18935

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.18935
License (as archived)
CC0-1.0
Archived
2026-10-15T13:49:38.877Z
SHA-256 of the archived text
866a331f200a406ae6ca17de68fdce7a147334c87c71f61783e724a30755aa92
Stable URL of this page
https://metraipolis.com/documento/dc_debdee463029

How to cite

Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs — arXiv:2609.18935 — arXiv (en), https://arxiv.org/abs/2609.18935 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_debdee463029 (Metraipolis archive dc_debdee463029, sha256 866a331f200a406ae6ca17de68fdce7a147334c87c71f61783e724a30755aa92)
BibTeX
@misc{dcdebdee463029,
  title = {Long-Lived Characters, Local Inference: Incremental Memory Maintenance for Game NPCs — arXiv:2609.18935},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.18935},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_debdee463029; sha256 866a331f200a406ae6ca17de68fdce7a147334c87c71f61783e724a30755aa92},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.18935. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

A game character should not have to reread its entire life before every conversation. For locally deployed language-model characters, however, revising a few memories can invalidate a long reusable prefix. The resulting preparation cost competes with both foreground dialogue and the maintenance of other characters. This matters especially when dialogue feeds game-defined actions and value judgments: a fluent but incorrect account of who owns an item, or whether a transfer has already happened, can corrupt the input to otherwise deterministic rules. We study incremental memory maintenance for long-lived game NPCs in a quantized Qwen hybrid recurrent-attention model. Our runtime removes superseded attention KV entries, computes replacement records at the true sequence tail, and preserves the continuing recurrent state and unchanged KV. Existing local experiments combine multi-update dialogue replays, fixed-input placement ablations, and attention diagnostics. Independent block composition weakens query-conditioned memory selection without a uniform chunk-initial attention collapse. True-tail updates preserve important current-state and historical bindings across eight scripted maintenance rounds; a placement case recovers the full-refill quantity in three reconstructions, while slot-preserving alternatives repeat a double-subtraction error. Attention-distribution proximity alone does not explain these semantic differences. The results motivate treating a character's inference state as a maintained, history-dependent resource, rather than only a disposable encoding of its latest memory text.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-17 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2609.18935 — 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 2026-10-15T13:48:24.353Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-17?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).