TimeOmni-1: Incentivizing Complex Reasoning with Time Series in Large Language Models

2025-09-29

AI Summary

Existing multimodal time-series datasets predominantly support superficial alignment and question-answering, lacking well-defined deep reasoning tasks and high-quality annotated data—hindering the development of Time-Series Reasoning Models (TSRMs). To address this, we propose TSR-Suite: the first atomic time-series reasoning task suite covering perception, extrapolation, and decision-making, accompanied by a unified training and evaluation framework and human-guided hierarchical annotation. Methodologically, we integrate task-mixed learning, custom reward functions, and multi-stage optimization, synergizing large language models with multimodal time-series modeling techniques. Our model, TimeOmni-1, achieves significant improvements in out-of-distribution generalization and response validity: causal discovery accuracy reaches 64.0% (a +28.1 percentage-point gain over GPT-4.1), and effective response rate for event-aware forecasting improves by over 6%.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Planning, Routing, and Scheduling: Model-Based Reasoning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Economics, Online Markets and Human Computation: Data quality aspects of human-annotated datasets

Abstract

Recent advances in multimodal time series learning underscore a paradigm shift from analytics centered on basic patterns toward advanced time series understanding and reasoning. However, existing multimodal time series datasets mostly remain at the level of surface alignment and question answering, without reaching the depth of genuine reasoning. The absence of well-defined tasks that genuinely require time series reasoning, along with the scarcity of high-quality data, has limited progress in building practical time series reasoning models (TSRMs). To this end, we introduce Time Series Reasoning Suite (TSR-Suite), which formalizes four atomic tasks that span three fundamental capabilities for reasoning with time series: (1) perception, acquired through scenario understanding and causality discovery; (2) extrapolation, realized via event-aware forecasting; and (3) decision-making, developed through deliberation over perception and extrapolation. TSR-Suite is the first comprehensive time series reasoning suite that supports not only thorough evaluation but also the data pipeline and training of TSRMs. It contains more than 23K samples, of which 2.3K are carefully curated through a human-guided hierarchical annotation process. Building on this foundation, we introduce TimeOmni-1, the first unified reasoning model designed to address diverse real-world problems demanding time series reasoning. The model is trained in multiple stages, integrating a mixture of task scenarios, novel reward functions, and tailored optimizations. Experiments show that TimeOmni-1 delivers strong out-of-distribution generalization across all tasks and achieves a high rate of valid responses. It significantly improves causality discovery accuracy (64.0% vs. 35.9% with GPT-4.1) and raises the valid response rate by over 6% compared to GPT-4.1 on the event-aware forecasting task.

Problem

• Addressing the lack of genuine time series reasoning in current multimodal datasets
• Developing a unified model for complex time series perception, extrapolation, and decision-making
• Overcoming limitations in causality discovery and event-aware forecasting accuracy

Innovation

• Introduces Time Series Reasoning Suite with four atomic tasks
• Proposes TimeOmni-1 model with multi-stage training approach
• Uses novel reward functions and tailored optimization methods