Metraipolis › archive › dc_6695e9cf485d

Learning What to Ask and What to Believe: Distilling a Variational Bayesian Agent into Language Models — arXiv:2611.01245

Source
arXiv (en)
Original
https://arxiv.org/abs/2611.01245
License (as archived)
CC0-1.0
Archived
2026-11-03T11:20:58.135Z
SHA-256 of the archived text
d648ed8658599fe1fd6ae5df594b082b05e9b7d725ef6bd4ba02a49308577ccf
Stable URL of this page
https://metraipolis.com/documento/dc_6695e9cf485d

How to cite

Learning What to Ask and What to Believe: Distilling a Variational Bayesian Agent into Language Models — arXiv:2611.01245 — arXiv (en), https://arxiv.org/abs/2611.01245 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_6695e9cf485d (Metraipolis archive dc_6695e9cf485d, sha256 d648ed8658599fe1fd6ae5df594b082b05e9b7d725ef6bd4ba02a49308577ccf)
BibTeX
@misc{dc6695e9cf485d,
  title = {Learning What to Ask and What to Believe: Distilling a Variational Bayesian Agent into Language Models — arXiv:2611.01245},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.01245},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_6695e9cf485d; sha256 d648ed8658599fe1fd6ae5df594b082b05e9b7d725ef6bd4ba02a49308577ccf},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.01245. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Interactive tasks such as eliciting a user's preferences or diagnosing a hidden mechanism require an agent to track its uncertainty over a latent state, ask informative questions and act on the answers. Language models have no explicit mechanism for this, whereas Bayesian information seeking provides one through posterior inference and expected information gain (EIG), and we ask whether it can be distilled into an 8B language model. Our teacher, a lightweight Bayesian agent with a sequential variational belief and EIG query selection, matches or approaches exact and particle references on three task families. We generate synthetic interactions with the teacher on episodes disjoint from the benchmark and fine-tune \Student{} to reproduce its queries and its belief state as a structured scratchpad. On three Multi-Turn Puzzles tasks the student raises Circuit Decoding from 0.00 to 0.93 success, lowers the normalized rank of the recommended film from 0.50 to 0.10 and cuts word guessing from 32 to 3.4 attempts, exceeding the strongest reported language models on the two comparable tasks. Exposing the belief state closes much of the gap where the decision is an implicit computation rather than a read-out of the interaction; in matched teacher contrasts, EIG supervision consistently outperforms random querying, whereas greater inference fidelity does not uniformly improve the distilled student; a 70B language-model teacher does not match the Bayesian teacher in downstream student performance, and the learned skill does not transfer to an unseen task family. Variational Bayesian information seeking can thus be amortized into a language model by transferring the belief state that drives its questions and decisions.

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno).
Categorie arXiv: Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.01245 — 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 2026-11-03T11:02:37.507Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).