Damien Ferbach, Courtney Paquette, Gauthier Gidel, Katie Everett, Elliot Paquette
In practice, the hyperparameters $(β_1, β_2)$ and weight-decay $λ$ in AdamW are typically kept at fixed values. Is there any reason to do otherwise? We show that for large-scale language model training, the answer is yes: by exploiting the power-law structure of language data, one can design time-varying schedules for $(β_1, β_2, λ)$ that deliver substantial performance gains. We study logarithmic-time scheduling, in which the optimizer's gradient memory horizon grows with training time. Although naive variants of this are unstable, we show that suitable damping mechanisms restore stability while preserving the benefits of longer memory. Based on this, we present ADANA, an AdamW-like optimizer that couples log-time schedules with explicit damping to balance stability and performance. We empirically evaluate ADANA across transformer scalings (45M to 2.6B parameters), comparing against AdamW, Muon, and AdEMAMix. When properly tuned, ADANA achieves up to 40% compute efficiency relative to a tuned AdamW, with gains that persist--and even improve--as model scale increases. We further show that similar benefits arise when applying logarithmic-time scheduling to AdEMAMix, and that logarithmic-time weight-decay alone can yield significant improvements. Finally, we present variants of ADANA that mitigate potential failure modes and improve robustness.
Quantitative mode stability for the wave equation on the Kerr-Newman spacetime
Risk-Aware Objective-Based Forecasting in Inertia Management
Chainalysis: Geography of Cryptocurrency 2023
Periodicity in Cryptocurrency Volatility and Liquidity
Impact of Geometric Uncertainty on the Computation of Abdominal Aortic Aneurysm Wall Strain
Simulation-based Bayesian inference with ameliorative learned summary statistics -- Part I