train_config.json

{
  "episode_dir": "./recorded_episodes",
  "output_dir": "./ppo_dynamic_checkpoints",
  "device": "cuda:0",
  "total_steps": 2000000,
  "n_rollout": 2048,
  "n_epochs": 4,
  "batch_size": 256,
  "lr": 0.0003,
  "gamma": 0.99,
  "gae_lambda": 0.95,
  "clip_eps": 0.2,
  "value_coef": 0.5,
  "entropy_coef": 0.01,
  "max_grad_norm": 0.5,
  "save_every": 50,
  "resume_from": null,
  "seed": 42,
  "dynamic_obstacles": true,
  "dynamic_config": "configs/dynamic_obstacles_medium.json",
  "max_dynamic_obstacles": null,
  "dynamic_min_speed": null,
  "dynamic_max_speed": null,
  "temporal_depth": true,
  "init_from": "./ppo_checkpoints2/policy_update_00976.pt",
  "hf_repo_id": null,
  "hf_token": null,
  "hf_best_only": false
}