The experiments are in papers_with_code/value_function and are generated by using the step_best_response_test - in TestWithSetupTabular is a option to comment loading of random strategies and instead load the cfr strategies
