behaviors:
  3DBall:
    trainer_type: ppo_transfer
    hyperparameters:
      batch_size: 64
      buffer_size: 12000
      learning_rate: 0.0003
      beta: 0.001
      epsilon: 0.2
      lambd: 0.99
      num_epoch: 3
      learning_rate_schedule: linear
      model_schedule: constant
      separate_model_train: true
      # separate_value_train: true
      encoder_layers: 1
      policy_layers: 1
      forward_layers: 1
      value_layers: 1
      feature_size: 16
      reuse_encoder: true
      in_epoch_alter: false
      in_batch_alter: false
      use_op_buffer: true
      use_var_predict: true
      with_prior: false
      predict_return: true
      use_bisim: false
      separate_value_net: false
    network_settings:
      normalize: true
      hidden_units: 128
      num_layers: 2
      vis_encode_type: simple
    reward_signals:
      extrinsic:
        gamma: 0.99
        strength: 1.0
    keep_checkpoints: 5
    max_steps: 2000000
    time_horizon: 1000
    summary_freq: 12000
    threaded: true