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

TL;DR

This paper introduces a comprehensive benchmarking protocol for neural network optimizers, evaluating end-to-end training efficiency and data-addition sensitivity across diverse tasks, revealing no universally superior optimizer.

Contribution

It proposes a novel benchmarking framework incorporating bandit hyperparameter tuning and data-shift sensitivity assessment, improving over previous random search methods.

Findings

No optimizer is best across all tasks.

Bandit hyperparameter tuning aligns better with human behavior.

The framework applies to multiple domains, including vision and NLP.

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.