Used SMAC in PyTorch to optimize the hyperparameters of a CNN which is trained on KMNIST dataset.
The optimizer used is Adam and the hyperparameters chosen for SMAC configuration space are learning rate and beta1.
First run python3 download_data.py
to download the KMNIST dataset.
Then run python3 kmnist_smac.py
to run the SMAC optimization.
The documentation for SMAC3 can be found here