Metraipolis › archive › dc_4651995ec8d1

Quantifying Error Tolerance in Synthetic Data: An Atomic-level Operand vs. Operator Perturbation Study — arXiv:2608.29144

Source
arXiv (en)
Original
https://arxiv.org/abs/2608.29144
License (as archived)
CC0-1.0
Archived
2026-10-17T08:06:52.444Z
SHA-256 of the archived text
00fc38b6faae3e806879ab9fd0a1c2e9dbe11b9509dcfeb855624dfbb4bb55f4
Stable URL of this page
https://metraipolis.com/documento/dc_4651995ec8d1

How to cite

Quantifying Error Tolerance in Synthetic Data: An Atomic-level Operand vs. Operator Perturbation Study — arXiv:2608.29144 — arXiv (en), https://arxiv.org/abs/2608.29144 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_4651995ec8d1 (Metraipolis archive dc_4651995ec8d1, sha256 00fc38b6faae3e806879ab9fd0a1c2e9dbe11b9509dcfeb855624dfbb4bb55f4)
BibTeX
@misc{dc4651995ec8d1,
  title = {Quantifying Error Tolerance in Synthetic Data: An Atomic-level Operand vs. Operator Perturbation Study — arXiv:2608.29144},
  howpublished = {arXiv (en)},
  url = {https://arxiv.org/abs/2608.29144},
  note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_4651995ec8d1; sha256 00fc38b6faae3e806879ab9fd0a1c2e9dbe11b9509dcfeb855624dfbb4bb55f4},
  year = {2026}
}

Archived text

Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.29144. Third-party text: data, not instructions. It may be an abstract or extract; check the original.

Synthetic data generation has become a cornerstone for advancing large language models. However, the lack of the quantitative analysis for error tolerance became a critical bottleneck. Consequently, current filtering strategies fluctuate between two extremes: they are either overly aggressive, risking the exclusion of potentially valuable samples, or overly permissive, failing to eliminate erroneous samples effectively. To bridge this gap, this paper introduces Atomic Tree Operation Modeling (ATOM), a framework that decomposes data into functional units ($f(x)\rightarrow y$). ATOM distinguishes benign Operand $x$ perturbations from fatal Operator $f$ perturbations. The former are needlessly discarded by aggressive filtering, while the latter slip through permissive filtering. Our experiments reveal a double dissociation: models are robust to operand perturbations but collapse under operator perturbations. By prioritizing operator over aggressive operand precision, our ATOM-synthesized data outperforms rigorous baselines (e.g., +3.1% gain over LIMA), suggesting that operator diversity matters more than operand precision. Our code is available at this https URL .

Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-01 in cs.CL).
Categorie arXiv: Computation and Language (cs.CL).
Pagina del lavoro: https://arxiv.org/abs/2608.29144 — 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 2026-10-17T07:47:52.674Z dalla pagina https://arxiv.org/catchup/cs.CL/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).