Metraipolis › archive › dc_abecbd2839cd

Mass Producing Training Data with Surprising Side Effects — arXiv:2611.06324

Source: arXiv (en)
Original: https://arxiv.org/abs/2611.06324
License (as archived): CC0-1.0
Archived: 2026-11-05T18:17:53.599Z
SHA-256 of the archived text: 61b288ee5d2e2f6545f3cc48a76a0374f0931b05e924b42b52f51caa5f8aac75
Stable URL of this page: https://metraipolis.com/documento/dc_abecbd2839cd

How to cite

Mass Producing Training Data with Surprising Side Effects — arXiv:2611.06324 — arXiv (en), https://arxiv.org/abs/2611.06324 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_abecbd2839cd (Metraipolis archive dc_abecbd2839cd, sha256 61b288ee5d2e2f6545f3cc48a76a0374f0931b05e924b42b52f51caa5f8aac75)

BibTeX

@misc{dcabecbd2839cd,
  title = {Mass Producing Training Data with Surprising Side Effects — arXiv:2611.06324},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2611.06324},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_abecbd2839cd; sha256 61b288ee5d2e2f6545f3cc48a76a0374f0931b05e924b42b52f51caa5f8aac75},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.06324. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Training data can have unexpected side effects on language model behavior outside the training domain: insecure code data can induce broad misalignment, and outdated bird names can induce a 19th-century persona. Understanding such generalization is hard because only dozens of isolated examples exist, each constructed by hand or discovered by accident. We propose automatically discovering such datasets with surprising side effects (DSSEs). We define three objectives for DSSEs: they should (1) elicit a target behavior, (2) read like ordinary post-training data, and (3) be surprising to human auditors reading the data. We operationalize each with LLM judges, and search for high-scoring datasets via evolution, reinforcement learning, and subliminal learning approaches. After re-scoring and filtering with held-out metrics, our pipeline produces a collection of 6,575 DSSEs across 8 behaviors, many of which generalize across model families and sizes. We find our automated metrics track ML researchers' judgments, and frontier models outperform humans at predicting side effects. These datasets form a testbed for studying generalization and lay the groundwork for predicting training side effects before model deployment. Explore them at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI); Machine Learning (cs.LG).
Pagina del lavoro: https://arxiv.org/abs/2611.06324 — 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 2026-11-05T18:12:49.054Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).