{
    "task": {
        "dataset": {
            "_target_": "datasets.make_dataset",
            "seq_len": 1,
            "pad_mode": "none",
            "val_ratio": 0.01,
            "obs_stack": 1,
            "action_chunk": 16,
            "to_memory": false,
            "dataset_id": "libero-90",
            "data_paths": "/opt/ml/input/data/training/libero_90/*.hdf5",
            "buffer_dir": "/tmp/buffers/libero_90/",
            "shape_meta": {
                "obs": {
                    "agentview_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    },
                    "eye_in_hand_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    }
                },
                "action": {
                    "shape": [
                        7
                    ]
                }
            },
            "batch_size": 200
        },
        "env": {
            "_target_": "environments.make_env",
            "seed": 42,
            "obs_stack": 1,
            "action_chunk": 16,
            "env_id": "libero-test",
            "hdf5_path": "/opt/ml/input/data/training/libero_90/KITCHEN_SCENE2_stack_the_middle_black_bowl_on_the_back_black_bowl_demo.hdf5",
            "shape_meta": {
                "obs": {
                    "agentview_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    },
                    "eye_in_hand_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    }
                },
                "action": {
                    "shape": [
                        7
                    ]
                }
            },
            "max_episode_steps": 1000
        }
    },
    "optimizer": {
        "_target_": "common.optimization.get_optimizer",
        "name": "AdamW",
        "model": "???",
        "lr": 0.0001,
        "weight_decay": 1e-06,
        "betas": [
            0.9,
            0.999
        ],
        "eps": 1e-08,
        "obs_encoder_lr_scale": 0.1
    },
    "scheduler": {
        "_target_": "common.optimization.get_scheduler",
        "name": "constant",
        "optimizer": "???",
        "num_warmup_steps": 20
    },
    "algo": "lmp-policy-ppo",
    "exp_id": "clipexpdiff0.05_ll0.01_s16v16_sig0.1-0.01",
    "seed": 42,
    "logdir": "/opt/ml/checkpoints/lmp-policy-ppo/libero_90/clipexpdiff0.05_ll0.01_s16v16_sig0.1-0.01/42",
    "resume": false,
    "use_amp": true,
    "compile": false,
    "pretrain_checkpoint_path": null,
    "num_steps": 5000000,
    "train_every": 3200,
    "num_train_epochs": 2,
    "eval_every": 128000,
    "save_every": 128000,
    "eval_rollout_every": 500000,
    "num_eval_rollouts": 20,
    "rollout_batch_size": 200,
    "train_batch_size": 64,
    "clip_grad_norm": 1.0,
    "free_nats_ratio": 0.05,
    "clip_range": 0.05,
    "alpha": 0.1,
    "kl_coef": 1.0,
    "ll_coef": 0.01,
    "model": {
        "_target_": "algorithms.latent_memory_palace.LatentMemoryPalacePolicy",
        "obs_encoder": {
            "_target_": "models.obs_encoders.SequenceObservationEncoder",
            "shape_meta": {
                "obs": {
                    "agentview_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    },
                    "eye_in_hand_rgb": {
                        "shape": [
                            128,
                            128,
                            3
                        ],
                        "type": "rgb"
                    }
                },
                "action": {
                    "shape": [
                        7
                    ]
                }
            },
            "embed_dim": 768,
            "num_timesteps": 1,
            "resize_shape": [
                240,
                240
            ],
            "crop_shape": [
                224,
                224
            ],
            "random_crop": false,
            "color_jitter": null,
            "vision_backbone": "pi05",
            "use_low_dim": true,
            "use_language": true,
            "pool_output": true,
            "max_language_length": 64,
            "num_film_layers": 2
        },
        "action_dim": 7,
        "action_chunk": 16,
        "seq_len": 16,
        "vocab_size": 16,
        "enc_embed_dim": 768,
        "enc_depth": 12,
        "enc_num_heads": 12,
        "enc_mlp_ratio": 4.0,
        "enc_num_registers": 1,
        "dec_embed_dim": 768,
        "dec_depth": 10,
        "dec_num_heads": 12,
        "dec_mlp_ratio": 4.0,
        "sigma_max": 0.1,
        "sigma_min": 0.01,
        "sigma_type": "learned_grid",
        "reg_type": "none",
        "reg_coef": 0
    }
}