T-LLM: Teaching Large Language Models to Forecast Time Series via Temporal Distillation

2026-02-02

AI Summary

Large language models (LLMs) struggle with time series forecasting due to their lack of explicit temporal modeling capabilities. This work proposes a temporal distillation framework that, for the first time, injects the predictive prowess of a lightweight teacher model into a general-purpose LLM through behavioral distillation. The teacher model integrates trend modeling and frequency-domain analysis to provide structured supervisory signals during training. Once trained, the LLM alone can perform end-to-end forecasting without requiring any additional inference modules. Evaluated across multiple benchmarks and infectious disease prediction tasks, the proposed approach consistently outperforms existing LLM-based forecasting methods under full-data, few-shot, and zero-shot settings.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Time series forecasting plays a critical role in decision-making across many real-world applications. Unlike data in vision and language domains, time series data is inherently tied to the evolution of underlying processes and can only accumulate as real-world time progresses, limiting the effectiveness of scale-driven pretraining alone. This time-bound constraint poses a challenge for enabling large language models (LLMs) to acquire forecasting capability, as existing approaches primarily rely on representation-level alignment or inference-time temporal modules rather than explicitly teaching forecasting behavior to the LLM. We propose T-LLM, a temporal distillation framework that equips general-purpose LLMs with time series forecasting capability by transferring predictive behavior from a lightweight temporal teacher during training. The teacher combines trend modeling and frequency-domain analysis to provide structured temporal supervision, and is removed entirely at inference, leaving the LLM as the sole forecasting model. Experiments on benchmark datasets and infectious disease forecasting tasks demonstrate that T-LLM consistently outperforms existing LLM-based forecasting methods under full-shot, few-shot, and zero-shot settings, while enabling a simple and efficient deployment pipeline.

Problem

• time series forecasting
• large language models
• temporal distillation
• forecasting capability
• time-bound data

Innovation

• Temporal Distillation
• Time Series Forecasting
• Large Language Models
• Teacher-Student Framework
• Frequency-Domain Analysis