📈
Production & Learning Effect
θ₀ (current)
—
θ₁ (after T)
—
Δθ
—
ΔY from θ (pure)
—
Total ΔY (Y₂−Y₀)
—
% explained by θ
Current MPR
—
Post MPR
—
How to read: Y₁ = more usage **without** learning. Y₂ = usage + **learning**. The stacked portion highlights the pure
gain from the data flywheel (Δθ). Try lowering λ or checking “Isolate learning”.
🧠
Learning Curve
Plots θ vs
x = λ·(0.03U + 0.04D). Points: current state (x₀) and after T cycles (x₁).