slm_lab/spec/experimental/dqn/ddqn_cartpole.json
{
"ddqn_boltzmann_cartpole": {
"agent": [{
"name": "DoubleDQN",
"algorithm": {
"name": "DoubleDQN",
"action_pdtype": "Categorical",
"action_policy": "boltzmann",
"explore_var_spec": {
"name": "linear_decay",
"start_val": 3.0,
"end_val": 1.0,
"start_step": 0,
"end_step": 2000,
},
"gamma": 0.99,
"training_batch_iter": 10,
"training_iter": 4,
"training_frequency": 8,
"training_start_step": 32
},
"memory": {
"name": "Replay",
"batch_size": 32,
"max_size": 10000,
"use_cer": true
},
"net": {
"type": "MLPNet",
"hid_layers": [64],
"hid_layers_activation": "selu",
"clip_grad_val": null,
"loss_spec": {
"name": "MSELoss"
},
"optim_spec": {
"name": "Adam",
"lr": 0.0002
},
"lr_scheduler_spec": {
"name": "StepLR",
"step_size": 500,
"gamma": 0.9,
},
"update_type": "polyak",
"update_frequency": 1,
"polyak_coef": 0,
"gpu": false
}
}],
"env": [{
"name": "CartPole-v0",
"max_t": null,
"max_frame": 400,
}],
"body": {
"product": "outer",
"num": 1
},
"meta": {
"distributed": false,
"eval_frequency": 1000,
"max_session": 4,
"max_trial": 100,
"search": "RandomSearch"
},
"search": {
"agent": [{
"algorithm": {
"explore_var_spec": {
"end_step__choice": [1000, 5000, 10000],
}
},
"net": {
"hid_layers_activation__choice": ["tanh", "relu", "selu"],
"optim_spec": {
"lr__uniform": [0.0001, 0.01]
}
}
}]
}
},
"ddqn_epsilon_greedy_cartpole": {
"agent": [{
"name": "DoubleDQN",
"algorithm": {
"name": "DoubleDQN",
"action_pdtype": "Argmax",
"action_policy": "epsilon_greedy",
"explore_var_spec": {
"name": "linear_decay",
"start_val": 1.0,
"end_val": 0.1,
"start_step": 0,
"end_step": 2000,
},
"gamma": 0.99,
"training_batch_iter": 8,
"training_iter": 4,
"training_frequency": 32,
"training_start_step": 10
},
"memory": {
"name": "Replay",
"batch_size": 32,
"max_size": 10000,
"use_cer": true
},
"net": {
"type": "MLPNet",
"hid_layers": [32],
"hid_layers_activation": "relu",
"clip_grad_val": null,
"loss_spec": {
"name": "MSELoss"
},
"optim_spec": {
"name": "Adam",
"lr": 0.02
},
"lr_scheduler_spec": {
"name": "StepLR",
"step_size": 500,
"gamma": 0.9,
},
"update_type": "polyak",
"update_frequency": 1,
"polyak_coef": 0.9,
"gpu": false
}
}],
"env": [{
"name": "CartPole-v0",
"max_t": null,
"max_frame": 250,
}],
"body": {
"product": "outer",
"num": 1
},
"meta": {
"distributed": false,
"eval_frequency": 1000,
"max_session": 4,
"max_trial": 100,
"search": "RandomSearch"
},
"search": {
"agent": [{
"algorithm": {
"explore_var_spec": {
"end_step__choice": [1000, 3000, 5000, 10000],
}
},
"net": {
"hid_layers__choice": [
[32],
[64],
[32, 16],
[64, 32]
],
"hid_layers_activation__choice": ["sigmoid", "relu", "tanh"],
"optim_spec": {
"lr__uniform": [0.0001, 0.2]
}
}
}]
}
},
"ddrqn_boltzmann_cartpole": {
"agent": [{
"name": "DDRQN",
"algorithm": {
"name": "DoubleDQN",
"action_pdtype": "Categorical",
"action_policy": "boltzmann",
"explore_var_spec": {
"name": "linear_decay",
"start_val": 3.0,
"end_val": 1.0,
"start_step": 0,
"end_step": 2000,
},
"gamma": 0.99,
"training_batch_iter": 8,
"training_iter": 4,
"training_frequency": 32,
"training_start_step": 10
},
"memory": {
"name": "Replay",
"batch_size": 32,
"max_size": 10000,
"use_cer": true
},
"net": {
"type": "RecurrentNet",
"cell_type": "GRU",
"fc_hid_layers": [],
"hid_layers_activation": "relu",
"rnn_hidden_size": 32,
"rnn_num_layers": 1,
"seq_len": 4,
"clip_grad_val": null,
"loss_spec": {
"name": "MSELoss"
},
"optim_spec": {
"name": "Adam",
"lr": 0.02
},
"lr_scheduler_spec": {
"name": "StepLR",
"step_size": 500,
"gamma": 0.9,
},
"update_type": "polyak",
"update_frequency": 1,
"polyak_coef": 0.9,
"gpu": false
}
}],
"env": [{
"name": "CartPole-v0",
"max_t": null,
"max_frame": 250,
}],
"body": {
"product": "outer",
"num": 1
},
"meta": {
"distributed": false,
"eval_frequency": 1000,
"max_session": 4,
"max_trial": 100,
"search": "RandomSearch"
},
"search": {
"agent": [{
"algorithm": {
"explore_var_spec": {
"end_step__choice": [1000, 3000, 5000, 10000],
}
},
"net": {
"hid_layers__choice": [
[32],
[64],
[32, 16],
[64, 32]
],
"hid_layers_activation__choice": ["sigmoid", "relu", "tanh"],
"optim_spec": {
"lr__uniform": [0.0001, 0.2]
}
}
}]
}
},
"ddrqn_epsilon_greedy_cartpole": {
"agent": [{
"name": "DDRQN",
"algorithm": {
"name": "DoubleDQN",
"action_pdtype": "Argmax",
"action_policy": "epsilon_greedy",
"explore_var_spec": {
"name": "linear_decay",
"start_val": 1.0,
"end_val": 0.1,
"start_step": 0,
"end_step": 2000,
},
"gamma": 0.99,
"training_batch_iter": 8,
"training_iter": 4,
"training_frequency": 32,
"training_start_step": 10
},
"memory": {
"name": "Replay",
"batch_size": 32,
"max_size": 10000,
"use_cer": true
},
"net": {
"type": "RecurrentNet",
"cell_type": "GRU",
"fc_hid_layers": [],
"hid_layers_activation": "relu",
"rnn_hidden_size": 32,
"rnn_num_layers": 1,
"seq_len": 4,
"clip_grad_val": null,
"loss_spec": {
"name": "MSELoss"
},
"optim_spec": {
"name": "Adam",
"lr": 0.02
},
"lr_scheduler_spec": {
"name": "StepLR",
"step_size": 500,
"gamma": 0.9,
},
"update_type": "polyak",
"update_frequency": 1,
"polyak_coef": 0.9,
"gpu": false
}
}],
"env": [{
"name": "CartPole-v0",
"max_t": null,
"max_frame": 250,
}],
"body": {
"product": "outer",
"num": 1
},
"meta": {
"distributed": false,
"eval_frequency": 1000,
"max_session": 4,
"max_trial": 100,
"search": "RandomSearch"
},
"search": {
"agent": [{
"algorithm": {
"explore_var_spec": {
"end_step__choice": [1000, 3000, 5000, 10000],
}
},
"net": {
"hid_layers__choice": [
[32],
[64],
[32, 16],
[64, 32]
],
"hid_layers_activation__choice": ["sigmoid", "relu", "tanh"],
"optim_spec": {
"lr__uniform": [0.0001, 0.2]
}
}
}]
}
},
}