[2603.16105] Frequency Matters: Fast Model-Agnostic Data Curation for Pruning and Quantization

Computer Science > Computation and Language arXiv:2603.16105 (cs) [Submitted on 17 Mar 2026 (v1), last revised 7 Apr 2026 (this version, v2)] Title:Frequency Matters: Fast Model-Agnostic Data Curation for Pruning and Quantization Authors:Francesco Pio Monaco, Elia Cunegatti, Flavio Vella, Giovanni Iacca View a PDF of the paper titled Frequency Matters: Fast Model-Agnostic Data Curation for Pruning and Quantization, by Francesco Pio Monaco and 3 other authors View PDF HTML (experimental) Abstract:Post-training model compression is essential for enhancing the portability of Large Language Models (LLMs) while preserving their performance. While several compression approaches have been proposed, less emphasis has been placed on selecting the most suitable set of data (the so-called \emph{calibration data}) for finding the compressed model configuration. The choice of calibration data is a critical step in preserving model capabilities both intra- and inter-tasks. In this work, we address the challenge of identifying high-performance calibration sets for both pruning and quantization by analyzing intrinsic data properties rather than model-specific signals. We introduce \texttt{\textbf{ZipCal}}, a model-agnostic data curation strategy that maximizes lexical diversity based on Zipfian power laws. Experiments demonstrate that our method consistently outperforms standard uniform random sampling across various pruning benchmarks. Notably, it also performs on par, in terms of downst...