ttnn.lerp

ttnn.lerp(input: ttnn.Tensor, end: ttnn.Tensor, weight: ttnn.Tensor or float, *, memory_config: ttnn.MemoryConfig = None, output_tensor: ttnn.Tensor = None) → None

Computes Lerp on input, end and weight and returns the tensor with the same layout as input

\[\mathrm{output\_tensor} = \verb|lerp|(\mathrm{input, end, weight})\]
Parameters:

  • input (ttnn.Tensor) – the input tensor with the starting points.

  • end (ttnn.Tensor) – the tensor with the ending points.

  • weight (ttnn.Tensor or float) – the weight for the interpolation formula.

Keyword Arguments:

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

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

Note

Supported dtypes and layouts:

Dtypes

Layouts

BFLOAT16, BFLOAT8_B, FLOAT32

TILE

bfloat8_b/bfloat4_b supports only on TILE_LAYOUT

end, weight tensors should have same dtype as input

output_tensor dtype must match input dtype, or be FLOAT32 when inputs are BFLOAT16, or be BFLOAT16 when inputs are FLOAT32

Example

# Create three tensors
tensor1 = ttnn.from_torch(
    torch.tensor([[1, 0], [1, 0]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device
)
tensor2 = ttnn.from_torch(
    torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device
)
tensor3 = ttnn.from_torch(
    torch.tensor([[1, 2], [3, 4]], dtype=torch.bfloat16), layout=ttnn.TILE_LAYOUT, device=device
)

# Perform the lerp operation
output = ttnn.lerp(tensor1, tensor2, tensor3)
logger.info(f"Lerp result: {output}")