Method __init__

ViNT class: uses a Transformer-based architecture to encode (current and past) visual observations and goals using an EfficientNet CNN, and predicts temporal distance and normalized actions in an embodiment-agnostic manner Args: context_size (int): how ma

(
        self,
        context_size: int = 5,
        len_traj_pred: Optional[int] = 5,
        learn_angle: Optional[bool] = True,
        obs_encoder: Optional[str] = "efficientnet-b0",
        obs_encoding_size: Optional[int] = 512,
        late_fusion: Optional[bool] = False,
        mha_num_attention_heads: Optional[int] = 2,
        mha_num_attention_layers: Optional[int] = 2,
        mha_ff_dim_factor: Optional[int] = 4,
    )