MMTS-BENCH: A Comprehensive Benchmark for Time Series Understanding and Reasoning Conference: ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 9–13, 2026; Jeju, Korea 83 CCS: Computing methodologies Natural language processing CCS: Computing methodologies Neural networks CCS: Gen
Yao Yin, Zhenyu Xiao, Musheng Li, Yiwen Liu, Sutong Nan, Yiting He, Ruiqi Wang, Zhenwei Zhang, Qingmin Liao, Yuantao Gu · 2026
Read MMTS-BENCH: A Comprehensive Benchmark for Time Series Understanding and Reasoning Conference: ACM SIGKDD Conference on Knowledge Discovery and Data Mining; August 9–13, 2026; Jeju, Korea 83 CCS: Computing methodologies Natural language processing CCS: Computing methodologies Neural networks CCS: Gen by Yao Yin, Zhenyu Xiao, Musheng Li, Yiwen Liu, Sutong Nan, Yiting He, Ruiqi Wang, Zhenwei Zhang, Qingmin Liao, Yuantao Gu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-08588.txt.
Description
Time series drive finance, health care, and cloud systems, yet benchmarks for language models on temporal tasks remain scattered. Yao Yin and nine co-authors introduce MMTS-BENCH, a multimodal benchmark organized by a hierarchy of time-series tasks. Their arXiv paper spans structure, features, temporal reasoning, matching, and cross-modal links.
The set holds 2,424 question-answer pairs across four subsets, built through a progressive real-world framework plus modular synthetic construction. Evaluations reveal four patterns: specialized time-series models lag general models on cross-domain transfer, local tasks are weaker than global ones, chain-of-thought and multimodal input help markedly, and backbone capability dominates encoder design. Readers get both a rigorous yardstick and clear directions for better temporal reasoning.
Catalog record
| CREATOR | Yao Yin, Zhenyu Xiao, Musheng Li, Yiwen Liu, Sutong Nan, Yiting He, Ruiqi Wang, Zhenwei Zhang, Qingmin Liao, Yuantao Gu |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 101 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08588