Showing Your Work Doesn't Always Work

TL;DR

This paper critically examines a popular method for reporting neural network results, revealing biases and limitations, and proposes an unbiased alternative supported by theoretical analysis and empirical evidence.

Contribution

It identifies biases in the existing estimator for reporting results and introduces an unbiased estimator with empirical validation.

Findings

Existing estimator is biased and error-prone.

Proposed unbiased estimator outperforms in simulations.

Confidence intervals are unreliable with the original method.

Abstract

In natural language processing, a recently popular line of work explores how to best report the experimental results of neural networks. One exemplar publication, titled "Show Your Work: Improved Reporting of Experimental Results," advocates for reporting the expected validation effectiveness of the best-tuned model, with respect to the computational budget. In the present work, we critically examine this paper. As far as statistical generalizability is concerned, we find unspoken pitfalls and caveats with this approach. We analytically show that their estimator is biased and uses error-prone assumptions. We find that the estimator favors negative errors and yields poor bootstrapped confidence intervals. We derive an unbiased alternative and bolster our claims with empirical evidence from statistical simulation. Our codebase is at http://github.com/castorini/meanmax.