Deriva — Derive the Algorithm

← Projectstensor-autograd / L3Computation Graph~20 min

one move: measure behavior

Build the scalar backward rule for z=a*b. Return both local responsibilities so a tiny autograd engine can accumulate them later. In the Tensor/Autograd Engine system, implement this level as a deterministic contract before composing it with the next service.

Required API

def backward_product(a, b):
    return {'da': ..., 'db': ...}

Behavior

assign the other operand as each local derivative
return named gradients
keep the rule independent of global state

Examples — tap to reveal

Constraints

Use only the Python standard library.
Return deterministic values so every run can be compared with the contract.
Do not mutate caller-owned inputs.

Opening…