From the paper, AFlow is trying to generate a optimized workflow from the val set of benchmarks, and then test the generated workflow on the test set. However, it seems you only have the code for optimizing on val set? Even if I uncomment optimizer.optimize("Test") in run.py, it is hardcoded to run the test set by using the round 1 workflow instead of final optimized workflow?
From the paper, AFlow is trying to generate a optimized workflow from the val set of benchmarks, and then test the generated workflow on the test set. However, it seems you only have the code for optimizing on val set? Even if I uncomment optimizer.optimize("Test") in run.py, it is hardcoded to run the test set by using the round 1 workflow instead of final optimized workflow?