ttnn.unsqueeze

ttnn.unsqueeze() → None

Returns a tensor unsqueezed at the specified dimension

Equivalent pytorch code:

input_tensor = torch.rand((1,1,256), dtype=torch.bfloat16)
output_tensor = torch.unsqueeze(input_tensor, 2) # tensor of shape (1,1,1,256), where at dimension 2 we added a new dim of size 1

:param * input_tensor: Input Tensor. :param * dim: Dim where we want to unsqueeze (add a new dimension of size 1)