Crate somatize_health

Expand description

Whether what happened is healthy. An opinion, and it says so.

The third of the three things observability splits into: the declaration drawn, the record of what happened, and a judgement about that record — with thresholds somebody chose, which is what makes it not a fact.

A diagnosis has to be reproducible from the stored record, without training again.

Which is why this crate has no dependencies and touches nothing: numbers in, Flags out. The statistics are measured where torch is, cross as numbers, and are judged here — so changing a threshold costs a scan, and an alarm you cannot re-ask is one you cannot argue with.

verdict asks whether a network is learning; leaning asks whether it is learning what you think, which no gradient will ever say.

It does not measure, and it does not decide when to look or how often: a verdict that quietly needed a particular cadence would be a threshold hiding in a schedule.

Dead and Saturated read the maximum and not the mean. A layer that dies one step in four is dead, and the average is what hides it. It is the original’s finding, and obvious only once somebody has been bitten.

Structs§

Contribution: What one input turned out to be worth.
Seen: The numbers a verdict is taken over.
Thresholds: The bounds a verdict is taken at.

Enums§

Flag: One thing that looks wrong with a node.

Functions§

leaning: What is wrong with what a model is leaning on. Empty is not a clean bill: with one input there is nothing to compare.
shares: Turns raw drops into shares, which is the only arithmetic here. Divided by the total and not the largest, so they add up to one. When nothing matters, every share is zero rather than a division by nothing, and IGNORED_INPUT fires on all of them — which is right.
verdict: Everything wrong with what was seen, in the order it is worth reading. Empty means nothing tripped, which is not the same as healthy — a metric nobody measured cannot flag, and Seen says which those were. What stops a run first comes first: a NaN makes every number below it meaningless.