OpenTSLM TeeMoE: A Unified Time-Series Language Model for Forecasting, Contextual Prediction, and Reasoning
OpenTSLM TeeMoE unifies forecasting, contextual prediction, and temporal reasoning with a LoRA mixture of experts.
OpenTSLM TeeMoE is a time-series language model that forecasts from observations, reasons over text and temporal patterns, and refines predictions from numerical specialists. Three low-rank experts cover forecast aggregation, native forecasting, and temporal analysis over a shared backbone, gated by a learned LoRA mixture-of-experts controller. It ranks among the top three on GIFT-Eval by mean MASE, Context is Key by RCRPS, and TimeSeriesExam by accuracy.
- Three frozen LoRA experts share one backbone.
- A learned controller weights experts per request.
- Top-three mean MASE rank on GIFT-Eval.
- Also strong on Context is Key and TimeSeriesExam.
- AI models
- OpenTSLM TeeMoE
From the original article
1 min read at arxiv.orgReal-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.
By Tony Chen · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.