params:
  seed: 42
  algo:
    name: a2c_continuous

  model:
    name: continuous_a2c_logstd

  network:
    name: actor_critic
    separate: False
    space:
      continuous:
        mu_activation: None
        sigma_activation: None
        mu_init:
          name: default
        sigma_init:
          name: const_initializer
          val: 1.2
        fixed_sigma: False
    mlp:
      units: [512, 256, 128]
      activation: relu
      d2rl: False
      initializer:
        name: default

  config:
    name: T1_Walking
    env_name: rlgym # Isaac Lab 包装器
    multi_gpu: False
    ppo: True
    mixed_precision: True
    normalize_input: True
    normalize_value: True
    value_bootstrap: True
    num_actors: 8192 # 同时训练的机器人数量
    reward_shaper:
      scale_value: 1.0
    normalize_advantage: True
    gamma: 0.96
    tau: 0.95
    learning_rate: 3e-4
    lr_schedule: adaptive
    kl_threshold: 0.008
    score_to_win: 20000
    max_epochs: 200
    save_best_after: 50
    save_frequency: 100
    grad_norm: 0.5
    entropy_coef: 0.008
    truncate_grads: True
    bounds_loss_coef: 0.001
    e_clip: 0.2
    horizon_length: 128
    minibatch_size: 8192
    mini_epochs: 4
    critic_coef: 1
    clip_value: True