LLaTiSA: Towards Difficulty-Stratified Time Series Reasoning from Visual Perception to Semantics
Fuente:
arXiv
Guardado en:
| Autores principales: | , , , , , , |
|---|---|
| Formato: | Preprint |
| Publicado: |
2026
|
| Materias: | |
| Acceso en línea: | |
| Etiquetas: |
Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
|
| _version_ | 1866910145885241344 |
|---|---|
| author | Ding, Yueyang Zhang, HaoPeng Dai, Rui Wang, Yi Zong, Tianyu Liu, Kaikui Chu, Xiangxiang |
| author_facet | Ding, Yueyang Zhang, HaoPeng Dai, Rui Wang, Yi Zong, Tianyu Liu, Kaikui Chu, Xiangxiang |
| contents | 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. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_17295 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | LLaTiSA: Towards Difficulty-Stratified Time Series Reasoning from Visual Perception to Semantics Ding, Yueyang Zhang, HaoPeng Dai, Rui Wang, Yi Zong, Tianyu Liu, Kaikui Chu, Xiangxiang Artificial Intelligence 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. |
| title | LLaTiSA: Towards Difficulty-Stratified Time Series Reasoning from Visual Perception to Semantics |
| topic | Artificial Intelligence |
| url | https://arxiv.org/abs/2604.17295 |