{
    "type": "fastwam",
    "n_obs_steps": 1,
    "input_features": {
        "observation.images.top": {
            "type": "VISUAL",
            "shape": [
                3,
                224,
                224
            ]
        },
        "observation.images.wrist": {
            "type": "VISUAL",
            "shape": [
                3,
                224,
                224
            ]
        },
        "observation.state": {
            "type": "STATE",
            "shape": [
                6
            ]
        }
    },
    "output_features": {
        "action": {
            "type": "ACTION",
            "shape": [
                6
            ]
        }
    },
    "device": "cuda",
    "use_amp": false,
    "use_peft": false,
    "push_to_hub": true,
    "repo_id": "maximellerbach/omx_multicubes_fastwam",
    "private": null,
    "tags": null,
    "license": null,
    "pretrained_path": null,
    "pretrained_revision": null,
    "action_dim": 6,
    "proprio_dim": 6,
    "action_horizon": 32,
    "n_action_steps": 32,
    "num_video_frames": 33,
    "action_video_freq_ratio": 4,
    "image_size": [
        224,
        448
    ],
    "context_len": 128,
    "model_id": "Wan-AI/Wan2.2-TI2V-5B",
    "tokenizer_model_id": "Wan-AI/Wan2.2-TI2V-5B",
    "base_model_id": "lerobot/fastwam_base",
    "tokenizer_max_len": 128,
    "load_text_encoder": true,
    "mot_checkpoint_mixed_attn": false,
    "torch_dtype": "bfloat16",
    "prompt_template": "A video recorded from a robot's point of view executing the following instruction: {task}",
    "num_inference_steps": 10,
    "inference_seed": 42,
    "rand_device": "cpu",
    "text_cfg_scale": 1.0,
    "negative_prompt": "",
    "sigma_shift": null,
    "tiled": false,
    "fp32_attention": true,
    "use_gradient_checkpointing": true,
    "freeze_video_expert": false,
    "toggle_action_dimensions": [],
    "video_scheduler": {
        "train_shift": 5.0,
        "infer_shift": 5.0,
        "num_train_timesteps": 1000
    },
    "action_scheduler": {
        "train_shift": 5.0,
        "infer_shift": 5.0,
        "num_train_timesteps": 1000
    },
    "loss": {
        "lambda_video": 1.0,
        "lambda_action": 1.0
    },
    "video_dit_config": {
        "patch_size": [
            1,
            2,
            2
        ],
        "in_dim": 48,
        "hidden_dim": 3072,
        "ffn_dim": 14336,
        "freq_dim": 256,
        "text_dim": 4096,
        "out_dim": 48,
        "num_heads": 24,
        "attn_head_dim": 128,
        "num_layers": 30,
        "eps": 1e-06,
        "seperated_timestep": true,
        "use_gradient_checkpointing": true,
        "video_attention_mask_mode": "first_frame_causal",
        "action_conditioned": false,
        "action_dim": 6,
        "action_group_causal_mask_mode": "group_diagonal",
        "fp32_attention": true
    },
    "action_dit_config": {
        "action_dim": 6,
        "hidden_dim": 1024,
        "ffn_dim": 4096,
        "num_heads": 24,
        "attn_head_dim": 128,
        "num_layers": 30,
        "text_dim": 4096,
        "freq_dim": 256,
        "eps": 1e-06,
        "use_gradient_checkpointing": true,
        "fp32_attention": true
    },
    "normalization_mapping": {
        "VISUAL": "IDENTITY",
        "STATE": "MEAN_STD",
        "ACTION": "MEAN_STD"
    },
    "optimizer_lr": 0.0001,
    "optimizer_weight_decay": 0.01
}