Morning Edition · Tuesday, September 15, 2026Published at 2:27 AM EDT · New York

Sakana AI trains 1,000-layer networks without backpropagation

The method accumulates per-layer constraint errors into a local Lagrange multiplier and lands within about two points of backpropagation accuracy on MNIST at that depth.

Sakana AI has published Augmented Lagrangian Predictive Coding, a training method that replaces the global backward pass with updates computed only from interactions between neighboring layers. Predictive coding has long been proposed as a…