We need to talk about random seeds

TL;DR

This paper discusses the proper and improper uses of random seeds in neural network training, highlighting risks and best practices to improve research reliability.

Contribution

It clarifies safe versus risky uses of random seeds in neural network experiments and provides an analysis of recent literature's practices.

Findings

Over 50% of recent publications have risky seed practices

Safe uses include hyperparameter search, ensembles, and sensitivity analysis

Risks involve fixed seeds for replicability and performance comparison

Abstract

Modern neural network libraries all take as a hyperparameter a random seed, typically used to determine the initial state of the model parameters. This opinion piece argues that there are some safe uses for random seeds: as part of the hyperparameter search to select a good model, creating an ensemble of several models, or measuring the sensitivity of the training algorithm to the random seed hyperparameter. It argues that some uses for random seeds are risky: using a fixed random seed for "replicability" and varying only the random seed to create score distributions for performance comparison. An analysis of 85 recent publications from the ACL Anthology finds that more than 50% contain risky uses of random seeds.