{
    "type": "ditflow",
    "n_obs_steps": 2,
    "input_features": {
        "observation.image.left_img": {
            "type": "VISUAL",
            "shape": [
                3,
                120,
                180
            ]
        },
        "observation.image.right_img": {
            "type": "VISUAL",
            "shape": [
                3,
                120,
                180
            ]
        },
        "observation.state": {
            "type": "STATE",
            "shape": [
                3
            ]
        }
    },
    "output_features": {
        "action": {
            "type": "ACTION",
            "shape": [
                3
            ]
        }
    },
    "device": "cuda",
    "use_amp": false,
    "push_to_hub": true,
    "repo_id": "MauoSama/dit_pushblock_unity",
    "private": null,
    "tags": null,
    "license": null,
    "pretrained_path": null,
    "horizon": 16,
    "n_action_steps": 8,
    "normalization_mapping": {
        "VISUAL": "MEAN_STD",
        "STATE": "MIN_MAX",
        "ACTION": "MIN_MAX"
    },
    "drop_n_last_frames": 7,
    "vision_backbone": "resnet18",
    "crop_shape": [
        84,
        84
    ],
    "crop_is_random": true,
    "pretrained_backbone_weights": null,
    "use_group_norm": true,
    "spatial_softmax_num_keypoints": 32,
    "use_separate_rgb_encoder_per_camera": true,
    "frequency_embedding_dim": 256,
    "hidden_dim": 512,
    "num_blocks": 6,
    "num_heads": 16,
    "dropout": 0.1,
    "dim_feedforward": 4096,
    "activation": "gelu",
    "training_noise_sampling": "uniform",
    "clip_sample": true,
    "clip_sample_range": 1.0,
    "num_inference_steps": 10,
    "do_mask_loss_for_padding": false,
    "optimizer_lr": 0.0001,
    "optimizer_betas": [
        0.95,
        0.999
    ],
    "optimizer_eps": 1e-08,
    "optimizer_weight_decay": 1e-06,
    "scheduler_name": "cosine",
    "scheduler_warmup_steps": 500
}