Follow the signal.

Run the engine. Inspect what changed. Trace every claim to evidence.

FULL EIDOS ENGINE · REAL DATA

Run an experiment

Choose data, review the settings, then launch the engine. You’ll get detection results, memory and regulation diagnostics, and downloadable evidence.

Engineering evaluationHeld-out data stays sealed

01Choose data & engineStart with the pinned example

02Check readinessReview and lock your settings

03Launch experimentOperator access required

04Inspect resultsMetrics, diagnostics & receipts

ENGINE OBSERVATORY

A trajectory of memory.

RECORDED EXAMPLE

Actual Torch engine · synthetic input · standard profile · seed 0. This example is separate from your experiment settings.

Inspect frame 549Viewing angle

Surprise z9.999

Distance from the engine's residual baseline, in robust scale units.

Memory gate χ0.037

Recorded familiarity gate: how strongly recognition inhibits learning.

Regulation T2.833

Internal noise-control parameter; this is not a physical temperature.

Three principal components preserve 82.2% of sampled state variance. Point color moves from mint to violet with sample order. Readouts use recorded frame 548; channel scales differ. This projection supports inspection, not proof of detection quality.

CIC-IDS2017 · network-flow example

A pinned file is ready below. Start with these settings to inspect engine behavior, or expand the source options to choose another dataset.

Engine profile

The established CPU profile. Reservoir prediction, episodic memory, adaptive learning and thermodynamic regulation are active.

Reservoir units: 256
Memory dimensions: 2,048
Time scales: 1
TraceSeal: Off

Dataset & advanced settings

DATASET LOOKUP

Pin the source

KAGGLE

DATASET HANDLEPINNED VERSIONENGINEERING SEEDEXACT FILE PATHLABEL COLUMNBENIGN LABELSROW ORDERORDER COLUMNMAX ROWSEXCLUDED COLUMNSEXPLICIT FEATURE COLUMNS optionalEXPECTED FILE SHA-256 optional first-run pin

File order is preserved, but chronology is unverified. This experiment can test engine behavior; it cannot establish operational detection performance.

WHAT THIS CAN ESTABLISH

Engineering behavior on pinned real observations, reproducibility receipts, causal preprocessing, and post-freeze evaluation metrics.

WHAT IT CANNOT ESTABLISH

Grand Proof acceptance, production safety, or held-out generalization. Those claims remain blocked.