Luca Zhou

@misc{zhou2026demystifyingmergeabilityinterpretableproperties,
  title={Demystifying Mergeability: Interpretable Properties to Predict Model Merging Success},
  author={Luca Zhou and Bo Zhao and Rose Yu and Emanuele Rodolà},
  year={2026},
  eprint={2601.22285},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2601.22285},
}

@inproceedings{zhou-etal-2026-cats,
title = "{C}a{TS}-Bench: Can Language Models Describe Time Series?",
author = "Zhou, Luca  and
  Yashwante, Pratham  and
  Fisher, Marshall  and
  Sampieri, Alessio  and
  Zhou, Zihao  and
  Galasso, Fabio  and
  Yu, Rose",
editor = "Liakata, Maria  and
  Moreira, Viviane P.  and
  Zhang, Jiajun  and
  Jurgens, David",
booktitle = "Findings of the {A}ssociation for {C}omputational {L}inguistics: {ACL} 2026",
month = jul,
year = "2026",
address = "San Diego, California, United States",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.findings-acl.1722/",
pages = "34479--34519",
ISBN = "979-8-89176-395-1",
abstract = "Time series captioning, the task of describing time series in natural language, requires numeric and temporal reasoning, trend interpretation, and contextual understanding. Existing benchmarks, however, often rely on fully synthetic or generic captions, and typically neglect metadata and visual representations. We introduce CaTS-Bench, a comprehensive benchmark for Context-aware Time Series reasoning across 11 diverse domains, centered on a gold-standard evaluation set of 1746 human-rewritten captions that measure how effectively models translate numeric trends into immediately interpretable narratives. To address the scarcity of human-annotated data, we also propose a scalable pipeline for generating high-fidelity synthetic captions, the quality of which we validate. We evaluate leading Vision-Language Models on our benchmark, revealing that even proprietary models struggle to capture numeric nuances in temporal descriptions, while finetuning open-source models on synthetic data yields substantial performance gains. Finally, we release a diagnostic suite of 910 multiple-choice questions and use tailored numeric metrics to gauge time-series-specific reasoning capabilities, establishing CaTS-Bench as a reliable foundation for grounded, multimodal text generation in numeric domains."
}

@misc{zhou2026hardjustunreacheddiagnosing,
  title={Hard or Just Unreached? Diagnosing the Sampling Blind Spot in Math-Reasoning Difficulty Estimation}, 
  author={Luca Zhou and Sajel Shah and Emanuele Rodolà and Roberto Dessì},
  year={2026},
  eprint={2606.19636},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2606.19636}, 
}

@InProceedings{pmlr-v322-zhou26a,
  title = {On Task Vectors and Gradients},
  author = {Zhou, Luca and Solombrino, Daniele and Crisostomi, Donato and Bucarelli, Maria Sofia and D'Inverno, Giuseppe Alessio and Silvestri, Fabrizio and Rodol\`{a}, Emanuele},
  booktitle = {Proceedings of UniReps: the Third Edition of the Workshop on Unifying Representations in Neural Models},
  pages = {398--417},
  year = {2026},
  editor = {Fumero, Marco and Domine, Clementine and L"ahner, Zorah and Cannistraci, Irene and Zhao, Bo and Williams, Alex},
  volume = {322},
  series = {Proceedings of Machine Learning Research},
  month = {06 Dec},
  publisher = {PMLR},
  pdf = {https://raw.githubusercontent.com/mlresearch/v322/main/assets/zhou26a/zhou26a.pdf},
  url = {https://proceedings.mlr.press/v322/zhou26a.html},
}

@inproceedings{zhou2025atm,
  title={Atm: Improving model merging by alternating tuning and merging},
  author={Zhou, Luca and Solombrino, Daniele and Crisostomi, Donato and Bucarelli, Maria Sofia and Silvestri, Fabrizio and Rodol{\`a}, Emanuele},
  booktitle={International Conference on Image Analysis and Processing},
  pages={192--201},
  year={2025},
  organization={Springer}
}

@misc{solombrino2026zeroshotquantizationweightspacearithmetic,
  title={Zero-Shot Quantization via Weight-Space Arithmetic},
  author={Daniele Solombrino and Antonio Andrea Gargiulo and Alessandro Zirilli and Luca Zhou and Adrian Robert Minut and Emanuele Rodolà},
  year={2026},
  eprint={2604.03420},
  archivePrefix={arXiv},
  primaryClass={cs.CV},
  url={https://arxiv.org/abs/2604.03420},
}

@misc{zhou2026tailshapeestimationllmevaluation,
  title={Tail-Shape Estimation in LLM Evaluation Is Fragile: A Protocol for Diagnosing False Positives},
  author={Luca Zhou},
  year={2026},
  eprint={2606.16511},
  archivePrefix={arXiv},
  primaryClass={cs.LG},
  url={https://arxiv.org/abs/2606.16511},
}