CoGenCast: A Coupled Autoregressive-Flow Generative Framework for Time Series Forecasting

TL;DR

CoGenCast introduces a hybrid time series forecasting framework that unifies semantic reasoning and continuous stochastic dynamics by reconfiguring a pre-trained decoder-only LLM into an encoder–decoder backbone and coupling it with a flow-matching mechanism. The encoder-decoder captures rich contextual semantics, while the flow-matching module models temporal evolution via learnable average velocity, enabling one-step generation. Extensive experiments across ten real-world datasets show state-of-the-art performance and robust generalization, with notable gains over both LLM-based and diffusion-based baselines and strong evidence for cross-domain training benefits. The work provides a practical, efficient approach to uncertainty-aware forecasting and highlights the value of integrating large-language-model priors with continuous probabilistic generation for time series tasks.

Abstract

Time series forecasting can be viewed as a generative problem that requires both semantic understanding over contextual conditions and stochastic modeling of continuous temporal dynamics. Existing approaches typically rely on either autoregressive large language models (LLMs) for semantic context modeling or diffusion-like models for continuous probabilistic generation. However, neither method alone can adequately model both aspects simultaneously. In this work, we propose CoGenCast, a hybrid generative framework that couples pre-trained LLMs with flow-matching mechanism for effective time series forecasting. Specifically, we reconfigure pre-trained decoder-only LLMs into a native forecasting encoder-decoder backbone by modifying only the attention topology, enabling bidirectional context encoding and causal representation generation. Building on this, a flow-matching mechanism is further integrated to model temporal evolution, capturing continuous stochastic dynamics conditioned on the autoregressively generated representation. Notably, CoGenCast naturally supports multimodal forecasting and cross-domain unified training. Extensive experiments on multiple benchmarks show that CoGenCast consistently outperforms previous compared baselines. Code is available at https://github.com/liuyaguo/_CoGenCast.

Figures (7)

• Figure 1: Overview of our proposed CoGenCast. Left (Training): We reconfigure decoder-only LLMs into an encoder--decoder backbone by attention-only modification, and perform continuous flow-matching mechanism conditioned on the LLM-generated representation. Right (Inference): Future patches are generated autoregressively and sampled via one-step flow-matching generation with low-latency.
• Figure 2: Performance comparison between in-domain and cross-domain training.The bar chart shows the MSE reduction achieved by cross-domain training across four representative benchmarks.
• Figure 3: Ablation study on context features. We evaluate the impact of removing domain knowdge, task instruction, statistics information, and the entire textual input on forecasting performance.
• Figure 4: Comparative analysis on the number of function evaluations (NFE). We compare the forecasting performance across 1, 2, and 3 sampling steps.
• Figure 5: Hyperparameter sensitivity analysis. (Left) Impact of varying patch sizes on MSE across different datasets. (Right) Performance comparison between linear and cosine noise schedules.