gelu¶

nvtripy.gelu(input: Tensor) → Tensor[source]¶

Applies a Gaussian Error Linear Unit (GELU) function to each element of the input tensor:

\(\text{gelu}(x) = x * \phi(x)\)

where \(\phi(x)\) is the Cumulative Distribution Function (CDF) of a Gaussian distribution.

Parameters:: input (Tensor) – The input tensor.
Returns:: A tensor of the same shape as the input.
Return type:: Tensor

INPUT REQUIREMENTS:: input.dtype is one of [float32, float16, bfloat16, int8]
OUTPUT GUARANTEES:: return[0].dtype == input.dtype

Example

input = tp.Tensor([1.0, 2.0, 3.0, 4.0])
output = tp.gelu(input)

Local Variables¶

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

>>> output
tensor([0.841345, 1.9545, 2.99595, 3.99987], dtype=float32, loc=gpu:0, shape=(4,))