Best values for training on Q-values so far are:
| Best Value So Far | Hyperparameter |
|---|---|
| 5 | depth |
| elu | activation |
| 0.0001 | lr_start |
| 160 | node_dim |
| 192 | n_output_dim |
| 192 | e_output_dim |
| 192 | emlp_output_dim |
| 160 | mlp_output_dim |
| 10 | tuner/epochs |
| 4 | tuner/initial_epoch |
| 1 | tuner/bracket |
| 1 | tuner/round |
Best val_loss So Far: 0.12021685391664505