| { | |
| "ONLINE PARAMETERS": "-------------------", | |
| "online_shared_parameters": { | |
| "replay_buffer_capacity": 1000000, | |
| "batch_size": 32, | |
| "update_horizon": 1, | |
| "gamma": 0.99, | |
| "learning_rate": 6.25e-05, | |
| "horizon": 27000, | |
| "n_epochs": 40, | |
| "n_training_steps_per_epoch": 250000, | |
| "n_initial_samples": 20000, | |
| "epsilon_end": 0.01, | |
| "epsilon_duration": 250000.0, | |
| "target_update_frequency": 8000, | |
| "update_to_data": 4.0, | |
| "features": [ | |
| 32, | |
| 64, | |
| 64, | |
| "Feature Size" | |
| ], | |
| "architecture_type": "cnn" | |
| }, | |
| "eaudedqn": { | |
| "n_networks": 5, | |
| "max_noise": 3.0, | |
| "max_speed": 0.01, | |
| "reset_optimizer": true | |
| }, | |
| "polyprunedqn": { | |
| "sparcity_start_step": 2000000, | |
| "sparcity_end_step": 8000000, | |
| "sparcity_update_freq": 4000, | |
| "final_sparsity": 0.95 | |
| }, | |
| "dqn": {}, | |
| "OFFLINE PARAMETERS": "-------------------", | |
| "offline_shared_parameters": { | |
| "features": [ | |
| 32, | |
| 64, | |
| 64, | |
| "Feature Size" | |
| ], | |
| "replay_buffer_capacity": 50000, | |
| "batch_size": 32, | |
| "update_horizon": 1, | |
| "gamma": 0.99, | |
| "learning_rate": 5e-05, | |
| "architecture_type": "cnn", | |
| "target_update_frequency": 2000, | |
| "n_buffers_to_load": 5, | |
| "n_epochs": 50, | |
| "n_fitting_steps": 62500 | |
| }, | |
| "eaudecql": { | |
| "n_networks": 5, | |
| "max_noise": 3.0, | |
| "max_speed": 0.01, | |
| "reset_optimizer": true, | |
| "alpha_cql": 0.1 | |
| }, | |
| "polyprunecql": { | |
| "sparcity_start_step": 625000, | |
| "sparcity_end_step": 2500000, | |
| "sparcity_update_freq": 1000, | |
| "final_sparsity": 0.95, | |
| "alpha_cql": 0.1 | |
| }, | |
| "cql": { | |
| "alpha_cql": 0.1 | |
| } | |
| } |