Search papers, labs, and topics across Lattice.
The paper formalizes Time Series Reasoning (TSR) with a four-level taxonomy and introduces HiTSR, a hierarchical dataset of 83k samples with Chain-of-Thought trajectories, to address the fragmented landscape of TSR benchmarks. They then propose LLaTiSA, a Vision-Language Model (VLM) that integrates visualized patterns and calibrated numerical tables to improve temporal perception. LLaTiSA, trained with a multi-stage curriculum fine-tuning strategy, demonstrates strong performance and out-of-distribution generalization on diverse TSR tasks.
Visualizing time series data and calibrating numerical tables unlocks surprisingly robust reasoning in Vision-Language Models, achieving strong generalization across diverse tasks.
Comprehensive understanding of time series remains a significant challenge for Large Language Models (LLMs). Current research is hindered by fragmented task definitions and benchmarks with inherent ambiguities, precluding rigorous evaluation and the development of unified Time Series Reasoning Models(TSRMs). To bridge this gap, we formalize Time Series Reasoning (TSR) via a four-level taxonomy of increasing cognitive complexity. We introduce HiTSR, a hierarchical time series reasoning dataset comprising 83k samples with diverse task combinations and verified Chain-of-Thought (CoT) trajectories. Leveraging HiTSR, we propose LLaTiSA, a strong TSRM that integrates visualized patterns with precision-calibrated numerical tables to enhance the temporal perception of Vision-Language Models (VLMs). Through a multi-stage curriculum fine-tuning strategy, LLaTiSA achieves superior performance and exhibits robust out-of-distribution generalization across diverse TSR tasks and real-world scenarios. Our code is available at https://github.com/RainingNovember/LLaTiSA.
