Closed seanlatias closed 3 years ago
Currently, we only allow users to access the shape property of a tensor. In this PR, we extend the use of shape to a tensor slice. We follow the same semantic as NumPy. For example,
shape
A = hcl.placeholder((2, 3, 4)) A.shape # (2, 3, 4) A[0].shape # (3, 4) A[0][1].shape # (4,) A[0][1][2].shape # ()
Currently, we only allow users to access the
shape
property of a tensor. In this PR, we extend the use ofshape
to a tensor slice. We follow the same semantic as NumPy. For example,