Dashboard
eigentau.ai

Strategy

Learned parameters and routing strategy configuration.

Network Regime
NEUTRAL
Auto-detected

Learned Parameters

decomposition_depth
4 2026-03-24

Optimal range 3-5. Beyond 5, synthesis quality degrades.

confidence_threshold
0.85 2026-03-24

Best win rate at confidence >= 0.85 (74% accuracy).

max_subnets_per_query
5 2026-03-23

Diminishing returns beyond 5 subnets per route.

ema_blend_factor
0.8 2026-03-22

80% current / 20% new prevents catastrophic forgetting.

fallback_timeout_ms
10000 2026-03-22

Give subnets 10s before falling back to direct inference.

min_quality_threshold
60 2026-03-21

Below 60% quality triggers re-route with different subnet combo.