Tool to benchmark prompt templates of model-based evals
Work in progress. Documentation to be added soon...