DataPraxis: Scaling Analytics Tasks via Verified Workspace Evolution

General-purpose agents are increasingly expected to perform professional analytics work, which requires sustained reasoning, iterative investigation, and reliable evidence. Training these capabilities requires diverse and challenging tasks.

人们日益期望通用智能体能够执行专业的分析工作，这需要持续的推理、迭代的调查以及可靠的证据。训练这些能力需要多样化且具有挑战性的任务。

However, generating tasks directly from raw data often overlooks deeper questions that emerge through analysis and provides limited evidence for verification. To address these limitations, we introduce DataPraxis, an analytics task synthesis framework based on verified workspace evolution.

然而，直接从原始数据生成任务往往会忽略分析过程中出现的更深层次问题，且提供的验证证据有限。为了解决这些局限性，我们引入了 DataPraxis，这是一个基于已验证工作空间演进的分析任务合成框架。

The framework progressively enriches workspaces with validated findings and artifacts, then inherits and combines these workspaces to uncover deeper questions. It constructs tasks from accumulated findings and their dependencies, verifying analytical results both during evolution and after task assembly.

该框架通过经过验证的发现和工件逐步丰富工作空间，然后继承并组合这些工作空间以发掘更深层次的问题。它利用积累的发现及其依赖关系构建任务，并在演进过程中和任务组装后对分析结果进行验证。

Using this framework, we construct DataPraxis-5K, a dataset of 5,000 verified trajectories spanning diverse domains and output formats, to train DataPraxis-27B and DataPraxis-35B-A3B. DataPraxis-27B achieves 55.11 on DABStep and 51.9 on JobBench, matching or surpassing frontier models such as GLM-5.2.

利用该框架，我们构建了 DataPraxis-5K，这是一个包含 5,000 条跨越不同领域和输出格式的已验证轨迹的数据集，用于训练 DataPraxis-27B 和 DataPraxis-35B-A3B。DataPraxis-27B 在 DABStep 上达到 55.11 分，在 JobBench 上达到 51.9 分，达到或超过了 GLM-5.2 等前沿模型。