works on
adam variant 1adaptive step control 1compute-matched evaluation 1gradient averaging 1runge-kutta optimizer 1

From the 1 of 1 linked paper with an AI index.

collaborators

1 paper

cs.LG2026

Adaptive Runge-Kutta Step Control Buys Training Loss, Not Generalization: An Honest Compute-Matched Study of RK-Adam Optimizers

Akhilesh Gogikar

The paper builds a Runge‑Kutta‑based variant of Adam and evaluates it under a strict compute‑matched budget, finding that the adaptive RK scheme does not improve generalization and…