{"ID":2844792,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2511.04973","arxiv_id":"2511.04973","title":"Less Is More: Generating Time Series with LLaMA-Style Autoregression in Simple Factorized Latent Spaces","abstract":"Generative models for multivariate time series are essential for data augmentation, simulation, and privacy preservation, yet current state-of-the-art diffusion-based approaches are slow and limited to fixed-length windows. We propose FAR-TS, a simple yet effective framework that combines disentangled factorization with an autoregressive Transformer over a discrete, quantized latent space to generate time series. Each time series is decomposed into a data-adaptive basis that captures static cross-channel correlations and temporal coefficients that are vector-quantized into discrete tokens. A LLaMA-style autoregressive Transformer then models these token sequences, enabling fast and controllable generation of sequences with arbitrary length. Owing to its streamlined design, FAR-TS achieves orders-of-magnitude faster generation than Diffusion-TS while preserving cross-channel correlations and an interpretable latent space, enabling high-quality and flexible time series synthesis.","short_abstract":"Generative models for multivariate time series are essential for data augmentation, simulation, and privacy preservation, yet current state-of-the-art diffusion-based approaches are slow and limited to fixed-length windows. We propose FAR-TS, a simple yet effective framework that combines disentangled factorization wit...","url_abs":"https://arxiv.org/abs/2511.04973","url_pdf":"https://arxiv.org/pdf/2511.04973v1","authors":"[\"Siyuan Li\",\"Yifan Sun\",\"Lei Cheng\",\"Lewen Wang\",\"Yang Liu\",\"Weiqing Liu\",\"Jianlong Li\",\"Jiang Bian\",\"Shikai Fang\"]","published":"2025-11-07T04:15:38Z","proceeding":"cs.LG","tasks":"[\"cs.LG\"]","methods":"[\"Diffusion Model\",\"Transformer\"]","has_code":false}