monad.ai

---

monad.ai / LEARNING_RATE

Variable: LEARNING_RATE

const LEARNING_RATE: 0.01 = 0.01

Defined in: kernel/adaptiveWeights.ts:24

Gradient step size (α) applied per forward result.

Small by design: Δweight = α × reward × contribution at α = 0.01 produces sub-1% shifts per step, keeping the weight trajectory smooth and preventing oscillation after a burst of unusual outcomes.