Metraipolis › archive › dc_14533cfceb02

PRISM: Promoting Sampling Diversity of Large Language Models with Implicit Stochastic Modulation — arXiv:2610.32101

Source
arXiv (en)
Original
https://arxiv.org/abs/2610.32101
License (as archived)
CC0-1.0
Archived
2026-10-26T20:11:48.449Z
SHA-256 of the archived text
7c160ad943d7ca19541a3c34d2bd59ccf8edbcb2453dd8e1061f213305a19053
Stable URL of this page
https://metraipolis.com/documento/dc_14533cfceb02

How to cite

PRISM: Promoting Sampling Diversity of Large Language Models with Implicit Stochastic Modulation — arXiv:2610.32101 — arXiv (en), https://arxiv.org/abs/2610.32101 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_14533cfceb02 (Metraipolis archive dc_14533cfceb02, sha256 7c160ad943d7ca19541a3c34d2bd59ccf8edbcb2453dd8e1061f213305a19053)
BibTeX
@misc{dc14533cfceb02,
  title = {PRISM: Promoting Sampling Diversity of Large Language Models with Implicit Stochastic Modulation — arXiv:2610.32101},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2610.32101},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_14533cfceb02; sha256 7c160ad943d7ca19541a3c34d2bd59ccf8edbcb2453dd8e1061f213305a19053},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.32101. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Large language models (LLMs) exhibit mode collapse, where their output distribution concentrates on familiar semantic modes. When diversity disappears, repeated sampling becomes semantic repetition rather than discovery, making rare but stronger solutions unlikely to surface and leaving self-evolution and continual learning difficult to sustain. Most remedies intervene only at inference time, shaping sam- ples without making diversity a model capability. We propose PRISM, a training-time framework that internalizes diversity within the model. PRISM combines (i) submodular curation that pairs each prompt with diverse-yet-correct multi-teacher solutions, and (ii) stochastic embedding modulation, which turns special-token perturbations into implicit signals for sampling distinct outputs. A theoretical analysis establishes the stability of this mechanism. On WritingPrompts and BookMIA, PRISM moves the quality-diversity frontier outward, achieving the best generation quality in all model-benchmark settings while maintaining the strongest or near-strongest semantic diversity across two open-weight model families. It also improves sampling efficiency on math and code benchmarks. Beyond inference, PRISM yields a stronger foundation for reinforcement learning, sustaining diverse rollouts and mitigating the diversity collapse commonly observed during RL training.

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/2610.32101 — 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 2026-10-26T19:49:03.383Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).