Update histograms with new vectors.

Update histograms with new vectors.

Take an action in the environment.

Take an action in the environment.

Optionally applied *per-dataset* transforms that happen at a frame level. Args: chunk_filter_fn (callable, optional): Filter functio

Ensure that array or tensor @x has size @size in dim @dim. Args: x (np.ndarray or torch.Tensor): input array or tensor size

Ensure that array or tensor @x has size @size in dim @dim. Args: x (np.ndarray or torch.Tensor): input array or tensor size

Convert rotations given as axis/angle to rotation matrices. Args: axis_angle: Rotations given as a vector in axis angle form,

Convert rotations given as axis/angle to quaternions. Args: axis_angle: Rotations given as a vector in axis angle form, a

将 base 坐标系下的矩阵转换到世界坐标系下

Calculate six camera positions and directions based on the scene's bounding box. Args: scene: The 3D scene Return

Center crop an image to match training data distribution. Args: image: Input image (PIL or numpy array) Returns: Image.

Recursively checks that elements in `values` match the TARGET_SPEC.

Check if two files are identical in content. Args: path1: Path to the first file path2: Path to the second file Returns

Validate input image format. Args: image: Image to check Raises: AssertionError: If image format is invalid

Check that the model contains the action un-normalization key.

Choose the best pose from the list of target poses. - target_lst: List of target poses.

Test the grasp pose function. - actor: The actor to be grasped. - arm_tag: The arm to be used for grasping, either "left" or

Yield n number of sequential chunks from l.

Yield n number of sequential chunks from l.

Yield n number of sequential chunks from l.

Yield n number of sequential chunks from l.

Yield n number of sequential chunks from l.

Yield n number of sequential chunks from l.

Dynamically import and instantiate the task implementation from the code_gen module. Args: task_name (str): Name of the task. R

Compose a sequence of transforms into a single transform.

lang_tokens: (batch_size, lang_len, lang_token_dim) lang_attn_mask: (batch_size, lang_len), a mask for valid language tokens,

Compute smoothened metrics from recent deques. Args: metrics_deques (dict): Dictionary of deques containing recent metrics. Ret

lang_cond: language conditional data, (batch_size, lang_len, hidden_size). lang_attn_mask: (batch_size, lang_len), a mask for valid l

Copy a on-disk zarr to in-memory compressed. Recommended

Copy a on-disk zarr to in-memory compressed. Recommended

Load to memory.

Load to memory.

Create an arrow pointing from start_point to end_point. Args: start_point: Starting coordinates of the arrow

Create a dataset for training.