arxiv.org
Verifiable evaluations of machine learning models using zkSNARKs
In a world of increasing closed-source commercial machine learning models, model evaluations from developers must be taken at face value. These benchmark results-whether over task accuracy, bias...
