A Generalized Framework for Population Based Training
KDD, pp. 1791.0-1799.0, 2019.
We show that the proposed Population Based Training framework is feasible for large scale deep neural network training
Population Based Training (PBT) is a recent approach that jointly optimizes neural network weights and hyperparameters which periodically copies weights of the best performers and mutates hyperparameters during training. Previous PBT implementations have been synchronized glass-box systems. We propose a general, black-box PBT framework th...More
PPT (Upload PPT)