ttnn.hardswish

ttnn.hardswish(input_tensor: ttnn.Tensor, *, scale: float | None = 0.16666667, shift: float | None = 0.5, memory_config: ttnn.MemoryConfig | None = None) → ttnn.Tensor

Performs hardswish function on input_tensor, scale, shift.

Parameters:

input_tensor (ttnn.Tensor) – the input tensor.

Keyword Arguments:

  • scale (float, optional) – Scale value. Defaults to 0.16666667.

  • shift (float, optional) – Shift value. Defaults to 0.5.

  • memory_config (ttnn.MemoryConfig, optional) – Memory configuration for the operation. Defaults to None.

Returns:

ttnn.Tensor – the output tensor.

Note

Supported dtypes, layouts, and ranks:

Dtypes

Layouts

Ranks

BFLOAT16, BFLOAT8_B

TILE

2, 3, 4

Example

>>> tensor = ttnn.from_torch(torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), dtype=ttnn.bfloat16, layout=ttnn.TILE_LAYOUT, device=device)
>>> output = ttnn.hardswish(tensor, scale = 0.16666667, shift = 0.5)