SeT-Diff: Towards Semantic Foundation Models for HPC Telemetry and Time-Series

SeT-Diff：面向高性能计算（HPC）遥测与时间序列的语义基础模型

Data centers and their compute nodes require accurate and flexible digital twins capable of modeling the complex interplay of workloads, environmental parameters, and physical metrics. 数据中心及其计算节点需要精确且灵活的数字孪生技术，以模拟工作负载、环境参数和物理指标之间复杂的相互作用。

Current machine learning approaches for HPC and its telemetry typically rely on a static subset of anonymous, fixed-position sensor variables tailored to single tasks. Consequently, these models become obsolete when target tasks change or sensor metrics vary. 目前针对高性能计算（HPC）及其遥测的机器学习方法，通常依赖于为特定任务定制的、匿名且位置固定的传感器变量静态子集。因此，当目标任务发生变化或传感器指标变动时，这些模型往往会失效。

We propose SeT-Diff, the first foundational model for compute node telemetry and time-series. Unlike rigid architectures, our diffusion-based approach conditions the generative process on each sensor’s semantic description, decoupling the system dynamics from the structure of the dataset. 我们提出了 SeT-Diff，这是首个针对计算节点遥测和时间序列的基础模型。与僵化的架构不同，我们基于扩散模型的方法将生成过程建立在每个传感器的语义描述之上，从而将系统动态与数据集结构解耦。

Experiments on a real-world supercomputer dataset demonstrate a Mean Absolute Error (MAE) of 0.0470 on reconstruction tasks. SeT-Diff exhibits zero-shot permutation stability, maintaining accuracy with negligible degradation even when sensors are shuffled. 在真实超级计算机数据集上的实验表明，该模型在重构任务中的平均绝对误差（MAE）为 0.0470。SeT-Diff 表现出了零样本置换稳定性，即使在传感器顺序被打乱的情况下，其精度也能保持且几乎没有下降。

A single pre-trained model effectively performs data imputation, forecasting, and virtual sensing - achieving a 0.033 MAE in thermal inference - making SeT-Diff an effective data-driven digital twin for HPC systems. 单一的预训练模型即可有效执行数据填补、预测和虚拟传感任务——在热推理中实现了 0.033 的 MAE——这使得 SeT-Diff 成为高性能计算系统的一种高效的数据驱动型数字孪生方案。