_tensor_round

tensor_round

Element-wise rounds the values of the tensor to nearest integer.

This method can be used with Numpy data:

n = np.array([[1.25, 4.5, 6], [4, 9.11, 16]])
b = fe.backend.tensor_round(n)  # [[1, 4, 6], [4, 9, 16]]

This method can be used with TensorFlow tensors:

t = tf.constant([[1.25, 4.5, 6], [4, 9.11, 16.9]])
b = fe.backend.tensor_round(t)  # [[1, 4, 6], [4, 9, 17]]

This method can be used with PyTorch tensors:

p = torch.tensor([[1.25, 4.5, 6], [4, 9.11, 16]])
b = fe.backend.tensor_round(p)  # [[1, 4, 6], [4, 9, 16]]

Parameters:

Name	Type	Description	Default
tensor	Tensor	The input tensor.	required

Returns:

Type	Description
Tensor	The rounded tensor.

Raises:

Type	Description
ValueError	If tensor is an unacceptable data type.

Source code in fastestimator/fastestimator/backend/_tensor_round.py

def tensor_round(tensor: Tensor) -> Tensor:
    """Element-wise rounds the values of the `tensor` to nearest integer.

    This method can be used with Numpy data:
    ```python
    n = np.array([[1.25, 4.5, 6], [4, 9.11, 16]])
    b = fe.backend.tensor_round(n)  # [[1, 4, 6], [4, 9, 16]]
    ```

    This method can be used with TensorFlow tensors:
    ```python
    t = tf.constant([[1.25, 4.5, 6], [4, 9.11, 16.9]])
    b = fe.backend.tensor_round(t)  # [[1, 4, 6], [4, 9, 17]]
    ```

    This method can be used with PyTorch tensors:
    ```python
    p = torch.tensor([[1.25, 4.5, 6], [4, 9.11, 16]])
    b = fe.backend.tensor_round(p)  # [[1, 4, 6], [4, 9, 16]]
    ```

    Args:
        tensor: The input tensor.

    Returns:
        The rounded `tensor`.

    Raises:
        ValueError: If `tensor` is an unacceptable data type.
    """
    if tf.is_tensor(tensor):
        return tf.round(tensor)
    elif isinstance(tensor, torch.Tensor):
        return torch.round(tensor)
    elif isinstance(tensor, np.ndarray):
        return np.round(tensor)
    else:
        raise ValueError("Unrecognized tensor type {}".format(type(tensor)))