config.json

{
  "env_paras": {
    "num_jobs": 10,
    "num_mas": 6,
    "batch_size": 1,
    "ope_feat_dim": 6,
    "ma_feat_dim": 3,
    "show_mode": "print",
    "valid_batch_size": 100
  },
  "model_paras": {
    "in_size_ma": 3,
    "out_size_ma": 8,
    "in_size_ope": 6,
    "out_size_ope": 8,
    "hidden_size_ope": 128,
    "num_heads": [
      1,
      1
    ],
    "dropout": 0.0,
    "n_latent_actor": 64,
    "n_latent_critic": 64,
    "n_hidden_actor": 3,
    "n_hidden_critic": 3,
    "action_dim": 1
  },
  "train_paras": {
    "lr": 0.0002,
    "betas": [
      0.9,
      0.999
    ],
    "gamma": 1.0,
    "K_epochs": 3,
    "eps_clip": 0.2,
    "A_coeff": 1.0,
    "vf_coeff": 0.5,
    "entropy_coeff": 0.01,
    "max_iterations": 1000,
    "save_timestep": 10,
    "update_timestep": 1,
    "viz": true,
    "viz_name": "Paper_1005",
    "minibatch_size": 512,
    "parallel_iter": 20
  },
  "test_paras": {
    "num_ins": 100,
    "rules": [
      "DRL"
    ],
    "sample": false,
    "num_sample": 1,
    "num_average": 1,
    "public_ins": true,
    "data_path": "1005test"
  },
  "TrainParas": {
    "max_episodes": 200001,
    "ep_max_step": 64,
    "random_seed": 0,
    "dataset": "1005test",
    "alpha": 0.00006,
    "gamma": 0.99,
    "tau": 0.005,
    "epsilon": 0.5,
    "eps_end": 0.0,
    "eps_dec": 1e-3,
    "max_size": 10000,
    "batch_size": 1024,
    "reward_type": 3,
    "render": false,
    "reset_mode": "Drl",
    "save_info": "actionmlp0503"
  },
  "LoadModelParas": {
    "ope_dim": 7,
    "gat_hidden_dim": 16,
    "gat_output_dim": 8,
    "hidden_layer_num": 3,
    "hidden_dim": 512,
    "load_ep": 93500,
    "load_info": "Recurrent",
    "load_reward_type": 3,
    "load_seed": 0
  }
}