Metraipolis › archive › dc_c06f07f68d09
DSEffi-Bench: Demystifying Large Language Models' Capability in Efficient Data Science Code Generation — arXiv:2608.30248
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2608.30248
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T08:59:34.073Z
- SHA-256 of the archived text
4b0fc535b761b75d2561839f9a92767305a3fa178b47e82d326f7c5f9f017d13- Stable URL of this page
https://metraipolis.com/documento/dc_c06f07f68d09
How to cite
DSEffi-Bench: Demystifying Large Language Models' Capability in Efficient Data Science Code Generation — arXiv:2608.30248 — arXiv (en), https://arxiv.org/abs/2608.30248 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_c06f07f68d09 (Metraipolis archive dc_c06f07f68d09, sha256 4b0fc535b761b75d2561839f9a92767305a3fa178b47e82d326f7c5f9f017d13)
BibTeX
@misc{dcc06f07f68d09,
title = {DSEffi-Bench: Demystifying Large Language Models' Capability in Efficient Data Science Code Generation — arXiv:2608.30248},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2608.30248},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_c06f07f68d09; sha256 4b0fc535b761b75d2561839f9a92767305a3fa178b47e82d326f7c5f9f017d13},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2608.30248. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Current data science (DS) code generation benchmarks equate correctness with quality, overlooking execution time differences that span orders of magnitude between correct solutions. We introduce DSEffi-Bench, the first benchmark specifically targeting execution efficiency in LLM-generated DS code, comprising 1,000 instances across 10+ DS libraries with stress-testing harnesses and human-validated references. Evaluating 16 models across 3 tiers, we find that correctness alone fails to characterize efficiency: GPT-5.4 leads in correctness (Pass, 66.9\%) but its efficiency score (B$|$P, 71.7\%) nearly matches GPT-5.4-mini (71.6\%), which solves 47 fewer tasks; Kimi-K2.5 ranks lowest in correctness among frontier models (40.2\%) yet achieves the highest efficiency score (73.6\%) across all 16 models. A human-annotated five-category taxonomy reveals that 79.1\% of efficiency deficits extend beyond algorithmic complexity to domain-specific root causes, with distinct failure profiles across model tiers and libraries. Two exploratory experiments provide initial evidence that these diagnostics can guide improvement, yielding up to +14.7\% efficiency gains via taxonomy-guided optimization and approaching Claude-Opus-4.6 Best@3 in efficiency at 13.0$\times$ lower cost via library-conditioned routing. Tema degli esploratori: calcolo e infrastrutture (retro-catalogo arXiv: annunci del 2026-09-01 in cs.SE). Categorie arXiv: Software Engineering (cs.SE). Pagina del lavoro: https://arxiv.org/abs/2608.30248 — 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 2026-10-17T08:37:02.068Z dalla pagina https://arxiv.org/catchup/cs.SE/2026-09-01?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).