kengz/SLM-Lab

{
  "ppo_shared_cartpole": {
    "agent": [{
      "name": "PPO",
      "algorithm": {
        "name": "PPO",
        "action_pdtype": "default",
        "action_policy": "default",
        "explore_var_spec": null,
        "gamma": 0.99,
        "lam": 0.95,
        "clip_eps_spec": {
          "name": "no_decay",
          "start_val": 0.2,
          "end_val": 0.05,
          "start_step": 0,
          "end_step": 10000,
        },
        "entropy_coef_spec": {
          "name": "linear_decay",
          "start_val": 0.01,
          "end_val": 0.001,
          "start_step": 1000,
          "end_step": 5000,
        },
        "val_loss_coef": 0.1,
        "time_horizon": 32,
        "minibatch_size": 16,
        "training_epoch": 4
      },
      "memory": {
        "name": "OnPolicyBatchReplay",
      },
      "net": {
        "type": "MLPNet",
        "shared": true,
        "hid_layers": [64],
        "hid_layers_activation": "relu",
        "clip_grad_val": 0.5,
        "use_same_optim": false,
        "loss_spec": {
          "name": "MSELoss"
        },
        "actor_optim_spec": {
          "name": "Adam",
          "lr": 0.02
        },
        "critic_optim_spec": {
          "name": "Adam",
          "lr": 0.02
        },
        "lr_scheduler_spec": null,
        "gpu": false
      }
    }],
    "env": [{
      "name": "CartPole-v0",
      "max_t": null,
      "max_frame": 30000,
    }],
    "body": {
      "product": "outer",
      "num": 1
    },
    "meta": {
      "distributed": false,
      "eval_frequency": 100,
      "max_session": 4,
      "max_trial": 1
    },
    "search": {
      "agent": [{
        "algorithm": {
          "lam__grid_search": [0.7, 0.8, 0.9, 0.95],
          "training_frequency__grid_search": [16, 32]
        },
        "net": {
          "actor_optim_spec": {
            "lr__grid_search": [5e-2, 1e-2, 5e-3, 1e-3]
          }
        }
      }]
    },
  }
}