full

nvtripy.full(shape: Sequence[int | DimensionSize], value: Tensor | Number, dtype: dtype = float32) → Tensor

Returns a tensor of the desired shape with all values set to the specified value.

Parameters:

• shape (Sequence[int | DimensionSize]) – The desired shape.

• value (Tensor | Number) – A scalar value to fill the resulting tensor.

• dtype (dtype) – The desired data type.

Returns:

A tensor of shape shape.

Return type:

Tensor

INPUT REQUIREMENTS:

• value.dtype != float8, and

• if value.dtype == int8, then dtype != bool, and

• dtype is one of [float32, float16, bfloat16, int8, int32, int64, bool]

OUTPUT GUARANTEES:

return[0].dtype == dtype

Example

output = tp.full(shape=[2, 3], value=2)

Local Variables

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