Reproduce “Examining Independence in Ensemble Sentiment Analysis: A Study on the Limits of Large Language Models Using the Condorcet Jury Theorem” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Baptiste Lefort, Eric Benhamou, Jean-Jacques Ohana, Beatrice Guez, David Saltiel, Thomas Jacquot
The authors test whether majority voting improves three-way financial-headline sentiment classification and whether model decisions are independent. Their proposed multi-class Condorcet framework expects gains when classifiers are competent and independent. In a Bloomberg headline experiment, voting combinations changed F-scores by only −0.01 to +0.01 relative to fine-tuned GPT-3.5, suggesting overlap without directly measuring it.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.