CAIN LatentGuard, explained for beginners
LatentGuard tells you to halt generation when the activation numbers you send it surge.
What it is
Some teams monitor a model's internal activations to catch unusual behaviour. LatentGuard is a simple screen over those numbers.
You send the statistics; a normal value is CLEAN; a surge past the threshold returns halt_generation = true with the pattern named.
An everyday comparison
A smoke alarm: you install the sensor, it sounds when the reading spikes.
How it works, in 4 steps
- Your system measures activations.
- It sends the statistics to LatentGuard.
- Normal: CLEAN.
- Surge: halt.
Try it now
Open the two live demo tools → Every button sends a real request to the live product and shows the real answer.
See it on the homepage: CAIN LatentGuard section →
Who it is for
ML teams running their own models who already measure activations.
FIRME AI CORPORATION · CAIN-42 is a Byzantine trust fabric. Numbers marked "our own check" were run by us, not by an independent party. Nothing on these pages is a certification.