Training Data Attribution via Approximate Unrolled Differentiation

Open in new window