KDD

FinSTaR: Towards Financial Reasoning with Time Series Reasoning Models

FinSTaR：面向时间序列推理模型的金融推理

Seunghan Lee, Jun Seo, Jaehoon Lee, Sungdong Yoo, Minjae Kim, Tae Yoon Lim, Dongwan Kang, Hwanil Choi, Soonyoung Lee, Wonbin Ahn

机构 * LG AI Research（LG人工智能研究） 纠错

AI总结 针对时间序列推理模型在金融领域的失效问题，提出基于2x2能力分类法的FinSTaR模型，通过Compute-in-CoT和Scenario-Aware CoT策略在FinTSR-Bench基准上达到78.9%平均准确率。

Comments EMNLP Industry track 2026, KDD Workshop on SciSoc Agents & LLMs 2026 (Oral Presentation)