Abstract:Deep learning (e.g., Transformer) has been widely and successfully used in multivariate time series forecasting (MTSF). Unlike existing methods that focus on training models from a single modal of time series input, large language models (LLMs) based MTSF methods with cross-modal text and time series input have recently shown great superiority, especially with limited temporal data. However, current LLM-based MTSF methods usually focus on adapting and fine-tuning LLMs, while neglecting the distribution discrepancy between textual and temporal input tokens, thus leading to sub-optimal performance. To address this issue, we propose a novel Cross-Modal LLM Fine-Tuning (CALF) framework for MTSF by reducing the distribution discrepancy between textual and temporal data, which mainly consists of the temporal target branch with temporal input and the textual source branch with aligned textual input. To reduce the distribution discrepancy, we develop the cross-modal match module to first align cross-modal input distributions. Additionally, to minimize the modality distribution gap in both feature and output spaces, feature regularization loss is developed to align the intermediate features between the two branches for better weight updates, while output consistency loss is introduced to allow the output representations of both branches to correspond effectively. Thanks to the modality alignment, CALF establishes state-of-the-art performance for both long-term and short-term forecasting tasks with low computational complexity, and exhibiting favorable few-shot and zero-shot abilities similar to that in LLMs. Code is available at \url{

Reprogramming Foundational Large Language Models(LLMs) for Enterprise Adoption for Spatio-Temporal Forecasting Applications: Unveiling a New Era in Copilot-Guided Cross-Modal Time Series Representation Learning

Time-LLM: Time Series Forecasting by Reprogramming Large Language Models

Advancing Enterprise Spatio-Temporal Forecasting Applications: Data Mining Meets Instruction Tuning of Language Models For Multi-modal Time Series Analysis in Low-Resource Settings

AutoTimes: Autoregressive Time Series Forecasters via Large Language Models

From News to Forecast: Integrating Event Analysis in LLM-Based Time Series Forecasting with Reflection

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

LLM4TS: Aligning Pre-Trained LLMs as Data-Efficient Time-Series Forecasters

Language Model Empowered Spatio-Temporal Forecasting via Physics-Aware Reprogramming

Effective LSTMs with Seasonal-Trend Decomposition and Adaptive Learning and Niching-Based Backtracking Search Algorithm for Time Series Forecasting

$\textbf{S}^2$IP-LLM: Semantic Space Informed Prompt Learning with LLM for Time Series Forecasting

TimeCMA: Towards LLM-Empowered Multivariate Time Series Forecasting via Cross-Modality Alignment

TS-TCD: Triplet-Level Cross-Modal Distillation for Time-Series Forecasting Using Large Language Models

Revisited Large Language Model for Time Series Analysis through Modality Alignment

LLMForecaster: Improving Seasonal Event Forecasts with Unstructured Textual Data

A spatio-temporal LSTM model to forecast across multiple temporal and spatial scales

CALF: Aligning LLMs for Time Series Forecasting via Cross-modal Fine-Tuning

LeRet: Language-Empowered Retentive Network for Time Series Forecasting

A Comprehensive Evaluation of Large Language Models on Temporal Event Forecasting

Rethinking Time Series Forecasting with LLMs via Nearest Neighbor Contrastive Learning

A federated large language model for long-term time series forecasting

How Can Large Language Models Understand Spatial-Temporal Data?