发表机构
Keiji AI（Keiji AI）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究通过线性探针审计可观测性基础模型Toto，发现其残差流中节奏和指标类型可恢复性优于原始输入模型，但可恢复性不等于实际使用，且外部基准零样本性能为负。
AI 中文摘要
线性探针可以表明一个标签可从模型隐藏状态中恢复，但无法表明这是否超出了输入本身已揭示的信息，或模型是否使用了该标签。我们审计了可观测性预测基础模型Toto，在可观测性指标基准（BOOM）上进行了五次序列不相交的重分割，将其冻结残差流上的线性探针与读取原始输入窗口的模型以及剥离了训练配置的Toto架构进行了比较。在每次重分割中，短与中节奏以及指标类型从Toto残差中的线性可恢复性均强于最强的原始窗口模型（宏F1分别为0.766对0.633和0.545对0.498）。领域几乎持平，而序列基数从原始窗口中恢复得更好。MOMENT-base显示出相关的节奏、指标类型和领域读数。可恢复性不等于使用：将Toto的残差与高突发供体的残差交换，按预期移动了未来突发性读数，但并未使预测比随机供体更持续地突发。一个在BOOM上训练的协调探针在测试的外部基准上具有负的零样本R^2。我们针对每个标签报告其相对于最强基线的结果。
英文摘要
A linear probe can show that a label is recoverable from a model's hidden states, but not whether that goes beyond what the input already reveals, or whether the model uses it. We audit Toto, an observability forecasting foundation model, on the Benchmark of Observability Metrics (BOOM) across five series-disjoint resplits, comparing linear probes on its frozen residual stream with models that read the raw input window and with Toto's architecture stripped of its trained configuration. Short-vs-medium cadence and metric type are more linearly recoverable from Toto's residuals than from the strongest raw-window model in every resplit (macro-F1 0.766 vs. 0.633 and 0.545 vs. 0.498). Domain is nearly tied, and series cardinality is recovered far better from the raw window. MOMENT-base shows related cadence, metric-type, and domain readouts. Recoverability is not use: exchanging Toto's residuals with those of high-burst donors moves a future-burstiness readout as intended but does not make forecasts consistently burstier than a randomized donor. A BOOM-trained coordination probe has negative zero-shot R^2 on the tested external benchmarks. We report each label against its strongest baseline.
CommentsAccepted to NeurIPS Main Conference '26. 25 pages
Journal refNeurIPS 2026
