Taming Pre-trained LLMs for Generalised Time Series Forecasting via Cross-modal Knowledge Distillation

arXiv:2403.07300v1 Announce Type: new
Abstract: Multivariate time series forecasting has recently gained great success with the rapid growth of deep learning models. However, existing approaches usually train models from scratch using limited temporal data, preventing their generalization. Recently, with the surge of the Large Language Models (LLMs), several works have attempted to introduce LLMs into time series forecasting. Despite promising results, these methods directly take time series as the input to LLMs, ignoring the inherent modality gap between temporal and …

arxiv cs.cl cs.lg distillation forecasting knowledge llms modal series time series time series forecasting type via