ttnn.sigmoid

ttnn.sigmoid(input_tensor: ttnn.Tensor, *, vector_mode: int = 4, mode: ttnn.SigmoidMode = SigmoidMode.Accurate, memory_config: ttnn.MemoryConfig = None, output_tensor: ttnn.Tensor = None) → ttnn.Tensor

Applies sigmoid to input_tensor element-wise.

\[\mathrm{output\_tensor}_i = \verb|sigmoid|(\mathrm{input\_tensor}_i)\]
Parameters:

input_tensor (ttnn.Tensor) – the input tensor.

Keyword Arguments:

  • vector_mode (int, optional) – Use vector mode to get better performance. Defaults to 4. Use 2 or 4 for different vector modes (2 -> Vector Mode C and 4 -> Vector Mode RC)”.

  • mode (ttnn.SigmoidMode, optional) – Select sigmoid mode to use. Defaults to SigmoidMode.Accurate.

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

  • output_tensor (ttnn.Tensor, optional) – preallocated output tensor. Defaults to None.

Returns:

ttnn.Tensor – the output tensor.

Note

Supported dtypes and layouts:

Dtypes

Layouts

BFLOAT16

TILE, ROW_MAJOR

Example

# Create a tensor with specific values
tensor = ttnn.from_torch(
    torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device
)

# Apply sigmoid activation function with vector mode
output = ttnn.sigmoid(tensor, vector_mode=4, mode=ttnn.SigmoidMode.FastApproximate)
logger.info(f"Sigmoid: {output}")