Unified model improves forecasting and temporal reasoning with text and numbers
OpenTSLM TeeMoE: A Unified Time-Series Language Model for Forecasting, Contextual Prediction, and Reasoning
Machine Learning
Summary
Many real-world problems with time-based data need models that can forecast future values, understand context from text, and reason about events over time. The authors created OpenTSLM TeeMoE, a model combining the strengths of numerical forecasters and language models. It uses multiple specialized parts that work together to provide accurate predictions and better analysis. This model performed very well on several established tests for forecasting and temporal understanding.
What this means in practice
- •For financial analysts: Generate accurate forecasts of market trends by integrating numerical models with contextual news and reports.
- •For energy grid operators: Improve demand forecasting using both observed time series and contextual information like weather forecasts and events.
Authors
Tony Chen, Timo Stoffregen, Maxwell Xu, Thomas Kaar, Martin Maritsch, Geremia Pompei, Nicolas Zumarraga, Robert Jakob, Paul Schmiedmayer, Patrick Langer, Juncheng Liu
Abstract
Real-world time-series applications increasingly require models that can handle time series forecasting, context-conditioned prediction, and language-based temporal reasoning. Yet current time-series foundation models remain fragmented across these capabilities: numerical specialists often provide the strongest forecasts, while language-based models offer broader contextual understanding and analysis. A central challenge is to unify these heterogeneous capabilities without reducing their individual performance. We introduce OpenTSLM TeeMoE, a generalist time-series language model that can forecast directly from observed time series, reason over textual context and temporal patterns, and synthesize and refine predictions from external numerical forecasting specialists. We independently train three low-rank experts for forecast aggregation, native forecasting, and temporal analysis over a shared backbone. A learned LoRA mixture-of-experts controller then weights their frozen parameter updates for each request. Our proposed model achieves strong performance on widely used benchmarks for time series forecasting, context-conditioned prediction, and language-based temporal reasoning, ranking among the top three on GIFT-Eval by mean MASE rank, Context is Key by RCRPS, and TimeSeriesExam by accuracy.