|TREND||DATASET||BEST METHOD||PAPER TITLE||PAPER||CODE||COMPARE|
Distributed training of deep learning models on large-scale training data is typically conducted with asynchronous stochastic optimization to maximize the rate of updates, at the cost of additional noise introduced from asynchrony.
While variance reduction methods have shown that reusing past gradients can be beneficial when there is a finite number of datapoints, they do not easily extend to the online setting.
The vast majority of successful deep neural networks are trained using variants of stochastic gradient descent (SGD) algorithms.
Partial warm restarts are also gaining popularity in gradient-based optimization to improve the rate of convergence in accelerated gradient schemes to deal with ill-conditioned functions.
Adaptive optimization methods such as AdaGrad, RMSProp and Adam have been proposed to achieve a rapid training process with an element-wise scaling term on learning rates.
In this paper, we first study a principled layerwise adaptation strategy to accelerate training of deep neural networks using large mini-batches.
#10 best model for Question Answering on SQuAD1.1 dev (F1 metric)
The learning rate warmup heuristic achieves remarkable success in stabilizing training, accelerating convergence and improving generalization for adaptive stochastic optimization algorithms like RMSprop and Adam.
We propose NovoGrad, an adaptive stochastic gradient descent method with layer-wise gradient normalization and decoupled weight decay.
The dynamic learning rate bounds are based on the exponential moving averages of the adaptive learning rates themselves, which smooth out unexpected large learning rates and stabilize the training of deep neural networks.