Metraipolis › archive › dc_bcbd41b873c1
ACE-Compression: An Ability-Centric Evaluation of Compressed Large Language Models — arXiv:2610.27913
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2610.27913
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-21T07:06:55.484Z
- SHA-256 of the archived text
db767a90a187381ab8810d52a6fc4bfe6616665bf523ef1e7a5636c9b647e482- Stable URL of this page
https://metraipolis.com/documento/dc_bcbd41b873c1
How to cite
ACE-Compression: An Ability-Centric Evaluation of Compressed Large Language Models — arXiv:2610.27913 — arXiv (en), https://arxiv.org/abs/2610.27913 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_bcbd41b873c1 (Metraipolis archive dc_bcbd41b873c1, sha256 db767a90a187381ab8810d52a6fc4bfe6616665bf523ef1e7a5636c9b647e482)
BibTeX
@misc{dcbcbd41b873c1,
title = {ACE-Compression: An Ability-Centric Evaluation of Compressed Large Language Models — arXiv:2610.27913},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2610.27913},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_bcbd41b873c1; sha256 db767a90a187381ab8810d52a6fc4bfe6616665bf523ef1e7a5636c9b647e482},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2610.27913. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Model compression makes large language models deployable, but today's evaluation tells an incomplete story. Standard practice measures retained task performance and stops there. This misses a critical fact: compression perturbs the distributed knowledge embedded across billions of weights, and task scores alone cannot reveal which underlying abilities survive and which quietly collapse. To solve this question, we introduce ACE-Compression, an ability-centric evaluation framework. Instead of asking only "does the compressed model still answer correctly?", we ask "which of its core abilities remain intact?" Our framework probes four deployment-facing dimensions: procedural reasoning and constraint execution, knowledge grounding and truthfulness, social cognition and affective understanding, and persona fidelity and stylistic consistency. Across more than 35 compressed checkpoints from six open-weight LLMs-covering quantization and pruning-we find that compression redistributes ability profiles. Coarse, local, and surface-level signals often persist, while fine-grained, long-tail, and context-dependent abilities become fragile. Quantization consistently outperforms pruning in retention, but the practical meaning of that gap depends entirely on which ability an application requires. Based on our findings, deployers can choose compression methods by the ability profile their application needs, not just by a leaderboard average. And algorithm designers gain diagnostic signals to build compression methods that preserve what matters. We believe that ACE-Compression turns compression from a black-box accuracy drop into an interpretable intervention-one we can measure, understand, and ultimately improve. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.AI, nuovi lavori del giorno). Categorie arXiv: Artificial Intelligence (cs.AI) ; Computation and Language (cs.CL). Pagina del lavoro: https://arxiv.org/abs/2610.27913 — 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 2026-10-21T07:05:43.454Z dalla pagina https://arxiv.org/list/cs.AI/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).