hyperparams/strat3_best_params.json

{
  "rl_lr": 0.00010813783435909571,
  "weight_decay": 7.7344821566468e-06,
  "dropout_p": 0.0599326836668414,
  "tmax": 5,
  "encoder_lr": 0.0002758206281092185,
  "head_lr": 0.0002908999360857405,
  "entropy_lr": 0.0007688106801474952,
  "smp_encoder_proj_dim": 192,
  "entropy_target_ratio": 0.43871283308838427,
  "entropy_alpha_init": 0.16391552721668934,
  "min_alpha": 0.0005693074767644616,
  "gae_lambda": 0.8944772161195532,
  "critic_loss_weight": 0.5212296337434775,
  "a3c_entropy_coeff": 0.003712624179040533,
  "asymmetric_reward_alpha": 0.5553304210317992,
  "boundary_reward_weight": 0.6095643339798968,
  "binary_reward_weight": 0.6016074139373169,
  "reward_boundary_focus": 0.33346118831249305,
  "reward_inaction_cost": 0.0490105049931256,
  "strategy3_rl_grad_clip_norm": 3.678930600883288,
  "refine_delta_small": 0.09791237813339448,
  "refine_delta_large": 0.20071320523192807,
  "threshold": 0.49683582808326887,
  "mc_dropout_p": 0.49497765893871026,
  "mc_n_passes": 4,
  "strategy3_decoder_unfreeze_warmup_fraction": 0.15,
  "strategy3_encoder_unfreeze_warmup_fraction": 0.45
}