@inproceedings{choi-etal-2025-rotate,
    title = "Rotate, Clip, and Partition: Towards {W}2{A}4{KV}4 Quantization by Integrating Rotation and Learnable Non-uniform Quantizer",
    author = "Choi, Euntae  and
      Song, Sumin  and
      Lim, Woosang  and
      Yoo, Sungjoo",
    editor = "Christodoulopoulos, Christos  and
      Chakraborty, Tanmoy  and
      Rose, Carolyn  and
      Peng, Violet",
    booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2025",
    month = nov,
    year = "2025",
    address = "Suzhou, China",
    publisher = "Association for Computational Linguistics",
    url = "https://preview.aclanthology.org/author-page-yu-wang-polytechnic/2025.findings-emnlp.400/",
    doi = "10.18653/v1/2025.findings-emnlp.400",
    pages = "7568--7590",
    ISBN = "979-8-89176-335-7",
    abstract = "We propose Rotate, Clip, and Partition (RCP), a Quantization-Aware Training (QAT) approach that first realizes extreme compression of LLMs with W2A4KV4 (2-bit weight, 4-bit activation, and 4-bit KV-cache) configuration. RCP integrates recent rotation techniques with a novel non-uniform weight quantizer design by theoretically and empirically analyzing the impact of rotation on the non-uniformity of weight distribution. Our weight quantizer, Learnable Direct Partitioning (LDP), introduces learnable parameters to directly learn non-uniform intervals jointly with LLM weights. We also present a GPU kernel supporting GEMV on non-uniform W2A4 as proof of concept. Experiments show that RCP can compress LLaMA-2-7B to W2A4KV4 with a loss of only 2.84 WikiText2 PPL and 5.29 times reduced memory footprint. Furthermore, RCP can quantize challenging mobile-targeted LLaMA-3.2 models and domain-specific WizardCoder-7B and MetaMath-7B with no critical problems such as convergence failure and repetition. Code is available at \url{https://github.com/songsm921/RCP}."
}