Function to break tensor along dim

1

There is a function in pytorch to break down a tensor into individual variables along a given dim. I can’t remember its name!
Something equivalent to x, y, z = a.moveaxis(d, 0).

2

You might be looking for torch.chunk or torch.split:

x = torch.randn(3, 2)
a, b, c = x.chunk(chunks=3, dim=-0)
print(a, b, c)
# tensor([[-1.4919,  2.4972]]) tensor([[ 0.4858, -1.2359]]) tensor([[0.4699, 0.3722]])