Prodigy: An Expeditiously Adaptive Parameter-Free Learner
This addresses the challenge of parameter tuning in machine learning optimization, offering a parameter-free adaptive method that is incremental but practical for broad applications.
The paper tackles the problem of estimating the learning rate in adaptive methods like AdaGrad and Adam by proposing Prodigy, an algorithm that provably estimates the distance to the solution for optimal learning rate setting, improving convergence rate by a factor of O(sqrt(log(D/d0))) over D-Adaptation and achieving test accuracy close to hand-tuned Adam across various benchmarks.
We consider the problem of estimating the learning rate in adaptive methods, such as AdaGrad and Adam. We propose Prodigy, an algorithm that provably estimates the distance to the solution $D$, which is needed to set the learning rate optimally. At its core, Prodigy is a modification of the D-Adaptation method for learning-rate-free learning. It improves upon the convergence rate of D-Adaptation by a factor of $O(\sqrt{\log(D/d_0)})$, where $d_0$ is the initial estimate of $D$. We test Prodigy on 12 common logistic-regression benchmark datasets, VGG11 and ResNet-50 training on CIFAR10, ViT training on Imagenet, LSTM training on IWSLT14, DLRM training on Criteo dataset, VarNet on Knee MRI dataset, as well as RoBERTa and GPT transformer training on BookWiki. Our experimental results show that our approach consistently outperforms D-Adaptation and reaches test accuracy values close to that of hand-tuned Adam.