[
    {
        "id": "osp-21647",
        "type": "article-journal",
        "title": "JARQ: Joint Alternating Refinement for Quantization",
        "author": [
            {
                "family": "Wang",
                "given": "Xinyu"
            },
            {
                "family": "Lyu",
                "given": "Sicheng"
            },
            {
                "family": "Chang",
                "given": "Xiao-Wen"
            }
        ],
        "URL": "https://omanscience.com/ar/articles/jarq-joint-alternating-refinement-for-quantization",
        "language": "en",
        "issued": {
            "date-parts": [
                [
                    2026
                ]
            ]
        },
        "abstract": "Group-wise post-training quantizers for large language models round weights onto a grid that is not refit to the resulting integer codes. We show that this leaves accuracy on the table: the best grid depends on the codes, input correlations couple the errors of different groups, and useful code changes often involve many codes at once. We propose JARQ , a plug-in refinement that starts from any group-wise quantizer and alternates a joint least-squares fit of all group scales with bounded Babai proposals that move many codes of a group together on the current grid. The problem is a bilinear box-constrained mixed-integer least-squares problem; the solver is backpropagation-free, does not increase the layer-wise objective under exact scale solves, and keeps the host's bit width, groups, zero points, and inference cost. Across Llama-2, Llama-3, and Qwen models with RTN, GPTQ, OmniQuant, and AWQ hosts, JARQ lowers perplexity in 90 of 96 comparisons, cuts three-bit RTN perplexity by up to 36%, raises mean multiple-choice accuracy in 23 of 24 configurations, and improves QEP, QuaRot, and OJBKQ outputs, at under a minute per 7B block."
    }
]