Table 1 cGAN hyperparameters. Parameters that were used for training are outlined in the table. Bolded specifications represent the parameters that were used to train the final model
Hyper-parameters | Specifications |
|---|---|
Batch size | 1, 2, 5, 10 |
Generator learning rate | 0.0000112, 0.000025, 0.00175, 0.0025 |
Discriminator learning rate | 0.0000045,0.0000075,0.00001, 0.0001 |
Loss | MSELoss, BCELoss, L1Loss |
Epochs | 50, 150, 250, 300, 750, 880, 1200, 1500 |
Leak value | 0.2 |