milton tensor-fusion

$\begin{aligned} model : θ \times ξ \to (χ \times ϕ) \to ψ \\ loss : ψ \times ψ \to ℓ \\ optimization : (θ \times ξ \times χ \to ψ) \to (ψ \times ψ \to ℓ) \to D \to θ \times λ \times L \\ inference : (θ \times κ \to ψ) \to θ \to (χ \times κ \times ζ) \to ψ \end{aligned}$