{
    "output_dir": null,
    "job_name": "so100_hilserl",
    "resume": false,
    "seed": 1000,
    "num_workers": 4,
    "batch_size": 256,
    "steps": 100000,
    "log_freq": 500,
    "save_checkpoint": true,
    "save_freq": 5000,
    "wandb": {
        "enable": true,
        "project": "so100_real",
        "disable_artifact": true
    },
    "dataset": {
        "repo_id": "lilkm/so100_gamepad_insert_high_quality",
        "use_imagenet_stats": false
    },
    "algorithm": {
        "type": "sac",
        "actor_lr": 3e-4,
        "critic_lr": 3e-4,
        "temperature_lr": 3e-4,
        "discount": 0.97,
        "use_backup_entropy": true,
        "critic_target_update_weight": 0.005,
        "num_critics": 2,
        "num_subsample_critics": null,
        "critic_network_kwargs": {
            "hidden_dims": [256, 256],
            "activate_final": true,
            "final_activation": null
        },
        "discrete_critic_network_kwargs": {
            "hidden_dims": [256, 256],
            "activate_final": true,
            "final_activation": null
        },
        "temperature_init": 0.01,
        "target_entropy": null,
        "utd_ratio": 2,
        "policy_update_freq": 1,
        "grad_clip_norm": 10.0,
        "use_torch_compile": false
    },
    "mixer": "online_offline",
    "online_ratio": 0.5,
    "policy": {
        "type": "gaussian_actor",
        "push_to_hub": false,
        "n_obs_steps": 1,
        "normalization_mapping": {
            "VISUAL": "MEAN_STD",
            "STATE": "MIN_MAX",
            "ENV": "MIN_MAX",
            "ACTION": "MIN_MAX"
        },
        "input_features": {
            "observation.images.top": {
                "type": "VISUAL",
                "shape": [3, 128, 128]
            },
            "observation.images.wrist": {
                "type": "VISUAL",
                "shape": [3, 128, 128]
            },
            "observation.state": {
                "type": "STATE",
                "shape": [18]
            }
        },
        "output_features": {
            "action": {
                "type": "ACTION",
                "shape": [3]
            }
        },
        "dataset_stats": {
            "observation.images.top": {
                "mean": [0.485, 0.456, 0.406],
                "std": [0.229, 0.224, 0.225]
            },
            "observation.images.wrist": {
                "mean": [0.485, 0.456, 0.406],
                "std": [0.229, 0.224, 0.225]
            },
            "observation.state": {
                "min": [-180, -180, -180, -180, -180, -180, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100, -100],
                "max": [180, 180, 180, 180, 180, 180, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100, 100]
            },
            "action": {
                "min": [-1.0, -1.0, -1.0],
                "max": [1.0, 1.0, 1.0]
            }
        },
        "num_discrete_actions": 3,
        "device": "cuda",
        "storage_device": "cuda",
        "use_amp": false,
        "vision_encoder_name": "lerobot/resnet10",
        "freeze_vision_encoder": true,
        "image_encoder_hidden_dim": 32,
        "shared_encoder": true,
        "online_steps": 1000000,
        "online_buffer_capacity": 30000,
        "offline_buffer_capacity": 20000,
        "online_step_before_learning": 100,
        "state_encoder_hidden_dim": 256,
        "latent_dim": 64,
        "actor_network_kwargs": {
            "hidden_dims": [256, 256],
            "activate_final": true
        },
        "policy_kwargs": {
            "use_tanh_squash": true,
            "std_min": -5,
            "std_max": 2,
            "init_final": 0.05
        },
        "actor_learner_config": {
            "learner_host": "127.0.0.1",
            "learner_port": 50051,
            "policy_parameters_push_frequency": 50
        },
        "concurrency": {
            "actor": "threads",
            "learner": "threads"
        },
        "repo_id": null
    },
    "env": {
        "type": "gym_manipulator",
        "name": "real_robot",
        "fps": 10,
        "task": "usb insert",
        "robot": {
            "type": "so100_follower",
            "port": "/dev/ttyACM0",
            "use_degrees": true,
            "id": "blue_follower",
            "cameras": {
                "top": {
                    "type": "opencv",
                    "index_or_path": "/dev/video2",
                    "height": 480,
                    "width": 640,
                    "fps": 30,
                    "fourcc": "MJPG"
                },
                "wrist": {
                    "type": "opencv",
                    "index_or_path": "/dev/video4",
                    "height": 480,
                    "width": 640,
                    "fps": 30,
                    "fourcc": "MJPG"
                }
            }
        },
        "teleop": {
            "type": "gamepad",
            "use_gripper": true
        },
        "processor": {
            "control_mode": "gamepad",
            "max_gripper_pos": 30,
            "observation": {
                "display_cameras": false,
                "add_joint_velocity_to_observation": true,
                "add_current_to_observation": true,
                "add_ee_pose_to_observation": false
            },
            "image_preprocessing": {
                "crop_params_dict": null,
                "resize_size": [128, 128]
            },
            "gripper": {
                "use_gripper": true,
                "gripper_penalty": -0.02
            },
            "reset": {
                "fixed_reset_joint_positions": [0.0, -20.0, 25.0, 90.0, 90.0, 30.0],
                "reset_time_s": 2.5,
                "control_time_s": 30.0,
                "terminate_on_success": true
            },
            "inverse_kinematics": {
                "urdf_path": "./SO101/so101_new_calib.urdf",
                "target_frame_name": "gripper_frame_link",
                "end_effector_bounds": {
                    "min": [0.115, -0.145, -0.0075],
                    "max": [0.26, 0.145, 0.1]
                },
                "end_effector_step_sizes": {
                    "x": 0.009,
                    "y": 0.009,
                    "z": 0.01
                }
            },
            "reward_classifier": {
                "pretrained_path": null,
                "success_threshold": 0.5,
                "success_reward": 1.0
            }
        },
        "features": {
            "observation.images.top": {
                "type": "VISUAL",
                "shape": [3, 128, 128]
            },
            "observation.images.wrist": {
                "type": "VISUAL",
                "shape": [3, 128, 128]
            },
            "observation.state": {
                "type": "STATE",
                "shape": [18]
            },
            "action": {
                "type": "ACTION",
                "shape": [3]
            }
        },
        "features_map": {
            "observation.images.top": "observation.images.top",
            "observation.images.wrist": "observation.images.wrist",
            "observation.state": "observation.state",
            "action": "action"
        }
    }
}