How much progress have we made in neural network training? A New Evaluation Protocol for Benchmarking Optimizers

Abstract

Many optimizers have been proposed for training deep neural networks, and they often have multiple hyperparameters, which make it tricky to benchmark their performance. In this work, we propose a new benchmarking protocol to evaluate both end-to-end efficiency (training a model from scratch without knowing the best hyperparameter) and data-addition training efficiency (the previously selected hyperparameters are used for periodically re-training the model with newly collected data). For end-to-end efficiency, unlike previous work that assumes random hyperparameter tuning, which over-emphasizes the tuning time, we propose to evaluate with a bandit hyperparameter tuning strategy. A human study is conducted to show that our evaluation protocol matches human tuning behavior better than the random search. For data-addition training, we propose a new protocol for assessing the hyperparameter sensitivity to data shift. We then apply the proposed benchmarking framework to 7 optimizers and various tasks, including computer vision, natural language processing, reinforcement learning, and graph mining. Our results show that there is no clear winner across all the tasks.

References in corpus (11)

• Auto-Encoding Variational Bayes
• Spectral Normalization for Generative Adversarial Networks
• On the Convergence of Adam and Beyond
• Cluster-GCN: An Efficient Algorithm for Training Deep and Large Graph Convolutional Networks
• On the Variance of the Adaptive Learning Rate and Beyond
• Large Batch Training of Convolutional Networks
• On Empirical Comparisons of Optimizers for Deep Learning
• Measuring the Effects of Data Parallelism on Neural Network Training
• The importance of better models in stochastic optimization
• Using a thousand optimization tasks to learn hyperparameter search strategies
• Minimum weight norm models do not always generalize well for over-parameterized problems