pad

nvtripy.pad(input: Tensor, pad: Sequence[Tuple[int | DimensionSize, int | DimensionSize]], mode: str = 'constant', value: int | float = 0) → Tensor

Pads the input tensor.

Parameters:

• input (Tensor) – [dtype=T1] The input tensor.

• pad (Sequence[Tuple[int | DimensionSize, int | DimensionSize]]) – A sequence of padding sizes of each dimension. Its length must be equal to the rank of input. Each element of pad is a tuple of integers or DimensionSize s (low, high), which represents the padding sizes before the lowest index and after the highest index at the corresponding dimension.

• mode (str) –

  The padding mode. Must be one of:

  • "constant": Pads with a constant value (default).

  • "reflect": Pads by reflecting the input tensor.

• value (int | float) – The padding value for “constant” mode. Has no effect for other modes.

Returns:

[dtype=T1] The padded tensor.

Return type:

Tensor

DATA TYPE CONSTRAINTS:

• T1: bool, float16, float32, int32, int64

Example: Constant Padding

input = tp.reshape(tp.arange(6, dtype=tp.float32), (2, 3))
output = tp.pad(input, [(1, 0), (0, 1)])

Local Variables

>>> input
tensor(
    [[0, 1, 2],
     [3, 4, 5]], 
    dtype=float32, loc=gpu:0, shape=(2, 3))

>>> output
tensor(
    [[0, 0, 0, 0],
     [0, 1, 2, 0],
     [3, 4, 5, 0]], 
    dtype=float32, loc=gpu:0, shape=(3, 4))

Example: Reflect Padding

input = tp.reshape(tp.arange(6, dtype=tp.float32), (2, 3))
output = tp.pad(input, [(1, 1), (1, 1)], mode="reflect")

Local Variables

>>> input
tensor(
    [[0, 1, 2],
     [3, 4, 5]], 
    dtype=float32, loc=gpu:0, shape=(2, 3))

>>> output
tensor(
    [[4, 3, 4, 5, 4],
     [1, 0, 1, 2, 1],
     [4, 3, 4, 5, 4],
     [1, 0, 1, 2, 1]], 
    dtype=float32, loc=gpu:0, shape=(4, 5))