Researcher questions evolutionary training routes, backs DNI for cases beyond backprop

AI researcher kalomaze dismissed evolutionary algorithm training routes as unappealing "Sakana-level nerd sniping," but argued that abandoning backprop makes sense when BPTT is infeasible, where modern DNI could reach otherwise unobservable training regimes.

2026-09-30 ~ 2026-09-30 · 2 related posts

• kalomaze calls ES/genetic-algorithm training pointless unless backprop is truly intractable — kalomaze · 2026-09-30
• kalomaze: Modern DNI May Approximate the Regimes Where Backprop Fails — kalomaze · 2026-09-30