Metraipolis › archive › dc_b47feb71dafa

Compiling Autoregressive LLM Control into Policy-Exact Parallel Inference for Hierarchical RAG — arXiv:2610.30664

Source: arXiv (en)
Original: https://arxiv.org/abs/2610.30664
License (as archived): CC0-1.0
Archived: 2026-10-23T15:20:27.667Z
SHA-256 of the archived text: 7c3bd40e73040abdc8ab7c3be0d157d082535511b76a7c63dd7cf7946931f416
Stable URL of this page: https://metraipolis.com/documento/dc_b47feb71dafa

How to cite

Compiling Autoregressive LLM Control into Policy-Exact Parallel Inference for Hierarchical RAG — arXiv:2610.30664 — arXiv (en), https://arxiv.org/abs/2610.30664 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_b47feb71dafa (Metraipolis archive dc_b47feb71dafa, sha256 7c3bd40e73040abdc8ab7c3be0d157d082535511b76a7c63dd7cf7946931f416)

BibTeX

@misc{dcb47feb71dafa,
  title = {Compiling Autoregressive LLM Control into Policy-Exact Parallel Inference for Hierarchical RAG — arXiv:2610.30664},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.30664},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_b47feb71dafa; sha256 7c3bd40e73040abdc8ab7c3be0d157d082535511b76a7c63dd7cf7946931f416},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.30664. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models increasingly act as controllers in structured decision pipelines, yet their decisions are still executed by autoregressive text decoding. Hierarchical RAG is one such setting: at each frontier of a document tree, an LLM commits to $b$ of $k$ sections under an input-dependent action grammar, executed as either failure-prone free-form JSON or reliable but strictly serial grammar-constrained generation. We show that this serialization is unnecessary for bounded-horizon control: the reachable future forms a shallow action-prefix tree, which we compile --- one root prefill plus one tree-causal probe evaluate every prefix-conditioned action distribution with serial decoding's tokens, positions, and causal ancestry. A prefix-state lemma makes compiled control policy-exact in exact arithmetic: the same policy with a two-forward critical path and zero output tokens. One further probe pass certifies the exact order-marginalized MAP set. Across four document-disjoint splits and two scales on the primary MetaX device, compiled control matches the serial policy on $84$--$89\%$ of questions --- every instrumented divergence a floating-point near-tie --- and cuts control latency by $35$--$47\%$ versus token-wise serial decoding and $6$--$7\%$ versus an idealized chunked engine; an H800 repetition preserves the ordering.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.CL, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Information Retrieval (cs.IR).
Pagina del lavoro: https://arxiv.org/abs/2610.30664 — 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 2026-10-23T15:09:38.321Z dalla pagina https://arxiv.org/list/cs.CL/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).