# Rethinking Calibration for High-Sparsity LLM Pruning: From Single Sources to Capability Coverage

arXiv: https://arxiv.org/abs/2606.03328 · HTML: https://mlsanity.com/paper/2606.03328/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 19 of its 26 references that are in the map.
- First posted on arXiv 2 Jun 2026.

## Builds on (most cited first, 19 of 19)

- [Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer](https://mlsanity.com/paper/1910.10683/) (2019)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge](https://mlsanity.com/paper/1803.05457/) (2018)
- [TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension](https://mlsanity.com/paper/1705.03551/) (2017)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformers](https://mlsanity.com/paper/2210.17323/) (2022)
- [Solving Quantitative Reasoning Problems with Language Models](https://mlsanity.com/paper/2206.14858/) (2022)
- [AWQ: Activation-aware Weight Quantization for LLM Compression and Acceleration](https://mlsanity.com/paper/2306.00978/) (2023)
- [SparseGPT: Massive Language Models Can Be Accurately Pruned in One-Shot](https://mlsanity.com/paper/2301.00774/) (2023)
- [The LAMBADA dataset: Word prediction requiring a broad discourse context](https://mlsanity.com/paper/1606.06031/) (2016)
- [CCNet: Extracting High Quality Monolingual Datasets from Web Crawl Data](https://mlsanity.com/paper/1911.00359/) (2019)
- [DataComp-LM: In search of the next generation of training sets for language models](https://mlsanity.com/paper/2406.11794/) (2024)
- [Compressing LLMs: The Truth is Rarely Pure and Never Simple](https://mlsanity.com/paper/2310.01382/) (2023)
- [On the Impact of Calibration Data in Post-training Quantization and Pruning](https://mlsanity.com/paper/2311.09755/) (2023)
- [Is C4 Dataset Optimal for Pruning? An Investigation of Calibration Data for LLM Pruning](https://mlsanity.com/paper/2410.07461/) (2024)
- [Self-calibration for Language Model Quantization and Pruning](https://mlsanity.com/paper/2410.17170/) (2024)
- [Do Compressed LLMs Forget Knowledge? An Experimental Study with Practical Implications](https://mlsanity.com/paper/2310.00867/) (2023)
- [On the Limitations of Language Targeted Pruning: Investigating the Calibration Language Impact in Multilingual LLM Pruning](https://mlsanity.com/paper/2408.14398/) (2024)
- [General Information Metrics for Improving AI Model Training Efficiency](https://mlsanity.com/paper/2501.02004/) (2025)

## Abstract (arXiv)

Post-training LLM pruning relies on a small calibration set, usually chosen heuristically and judged by WikiText perplexity or by mean accuracy on general-domain commonsense benchmarks. This single-axis view hides capability-specific effects: at 60% sparsity, strategies separated by 2.85 points in averaged commonsense accuracy diverge by 51.9 points in code retention, and across 15 sources no single one preserves all capabilities. We treat calibration as a capability-coverage problem. Balanced, a uniform four-way mixture under a fixed budget, tests the principle on real corpora: it outperforms every evaluated single source, and its margin over C4 grows from +4.6 points at 30% sparsity to +18.8 at 60%. The advantage holds across model scale and a second model family, and transfers to held-out benchmarks disjoint from calibration construction. When aligned corpora are unavailable, IGSP approximates that mixture from the base model and the evaluation taxonomy alone, outperforming prior self-generation methods by 7.23--12.79 points on the LLaMA main suite. On Qwen2.5-7B, IGSP exceeds Balanced on both the main suite and held-out benchmarks. Capability coverage is a first-order design variable for high-sparsity pruning.