Reproduce “Time-RA: Towards Time Series Reasoning for Anomaly Diagnosis with LLM Feedback” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yiyuan Yang, Zichuan Liu, Lei Song, Kai Ying, Zhiguang Wang, Tom Bamford, Svitlana Vyetrenko, Jiang Bian and 1 more

TIME-RA makes time-series anomaly detection a diagnostic task by generating a binary label, anomaly category, and textual rationale from a series, domain context, and optionally a plot. The authors introduce RATS40K, with 39,574 samples across 10 domains. On out-of-domain CATSv2, fine-tuned Qwen2.5-7B achieves F1 0.6606 for binary detection, below XGBoost’s 0.7826; diagnostic transfer is untested.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.