Datadog/ARFBench
收藏资源简介:
ARFBench(异常推理框架基准)是一个多模态时间序列推理基准,包含750个基于Datadog(一个领先的可观测性平台)收集的真实世界事件数据组成的问答对。这些数据覆盖了多个领域,包括应用程序使用、基础设施、网络、数据库和安全。每个问答对包括问题、任务类别、难度、选项、正确答案、查询组和插值标志。此外,每个唯一的时间序列都有两种不同的数据模态:时间序列数据和时序图。该基准旨在评估模型在软件事件响应中对多时间序列的推理能力。
ARFBench (Anomaly Reasoning Framework Benchmark) is a multimodal time-series reasoning benchmark consisting of 750 question-answer (QA) pairs composed from real-world incident data collected at Datadog, a leading observability platform. The data spans various domains including application usage, infrastructure, networking, database, and security. Each QA pair includes a question, task category, difficulty, options, correct answer, query group, and interpolation flags. Additionally, each unique time series has two associated modalities: time series data and time series plots. The benchmark is designed to evaluate the reasoning ability of models over multiple time series in software incident response.




