results_D512_teacher.json

{
  "mean_reward": 1000.0,
  "eval_mean_best": 744.6500244140625,
  "is_deterministic": true,
  "D": 512,
  "seed": 123,
  "algo": "Hybrid-HD-PPO",
  "env": "InvertedPendulum-v5",
  "beta_base": 0.7071067690849304,
  "beta_bands": null
}