Metraipolis › archive › dc_84d34fe2b936

Understanding Confabulation and Rethinking Reconstruction in Activation Explanations — arXiv:2609.33702

Source
arXiv (en)
Original
https://arxiv.org/abs/2609.33702
License (as archived)
CC0-1.0
Archived
2026-10-07T21:01:41.205Z
SHA-256 of the archived text
3c3f23b6ec7e8109b4753a26f7179ac770f870092d72c32c0c862eb88906a1e8
Stable URL of this page
https://metraipolis.com/documento/dc_84d34fe2b936

How to cite

Understanding Confabulation and Rethinking Reconstruction in Activation Explanations — arXiv:2609.33702 — arXiv (en), https://arxiv.org/abs/2609.33702 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_84d34fe2b936 (Metraipolis archive dc_84d34fe2b936, sha256 3c3f23b6ec7e8109b4753a26f7179ac770f870092d72c32c0c862eb88906a1e8)
BibTeX
@misc{dc84d34fe2b936,
  title = {Understanding Confabulation and Rethinking Reconstruction in Activation Explanations — arXiv:2609.33702},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2609.33702},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_84d34fe2b936; sha256 3c3f23b6ec7e8109b4753a26f7179ac770f870092d72c32c0c862eb88906a1e8},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.33702. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Natural Language Autoencoders (NLAs) produce unsupervised text explanations of a model's activations: a verbalizer describes an activation and a reconstructor learns to recover it from this text. Under the established point-reconstruction NLA training recipe, explanations become more useful for predicting model behavior while also increasingly introducing unsupported details and exhibiting writing defects. To assess these changes separately, we introduce a standardized evaluation framework for unstructured NLA explanations, measuring information recoverable from explanations, contextual support for their claims, and writing quality. To address confabulation and writing defects, we move beyond predicting a single activation: explanations can distinguish distributions of possible activations even when their means and optimal point-reconstruction rewards are identical. We introduce Flow-NLA, which models the distribution of activations compatible with an explanation and trains the verbalizer using a diffusion likelihood bound. Across Qwen, Gemma, and Apertus, this richer signal retains the utility gains of point reconstruction while curbing the growth of confabulation and writing defects, opening up a direction for improving activation-derived training to encourage more informative, supported, and readable explanations. Code and evaluation prompts will be made publicly available upon acceptance.

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-29 in cs.AI).
Categorie arXiv: Computation and Language (cs.CL) ; Artificial Intelligence (cs.AI).
Pagina del lavoro: https://arxiv.org/abs/2609.33702 — 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 2026-10-07T20:29:02.723Z dalla pagina https://arxiv.org/catchup/cs.AI/2026-09-29?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).