Avoid using legacy constructors Tensor.new_*()

[fritzo]

The following constructors are deprecated and are unsupported by the jit:

• Tensor.new_tensor()
• Tensor.new_empty()
• Tensor.new_zeros()
• Tensor.new_ones()
• Tensor.new_full()

Updated

• pyro core
• most of pyro.contrib
• examples/
• tests/infer/test_jit.py

Triaged (not updated)

• pyro.contrib.gp (I'll let @fehiepsi handle pyro.contrib.gp)
• tutorials
• tests

Tested

• refactoring is exercised by existing tests
• ran CUDA_TEST=1 pytest -vx tests/test_examples.py::test_cuda against torch==1.0.1

[neerajprad]

Not relevant to the cases that I see here, but I just discovered that torch.full doesn't generalize well in JIT when the fill value is scalar. Using empty_like with fill_ seems to work fine in those cases.

e.g.

>>> a = torch.tensor([0, 0])

>>> def fn1(x):
        return torch.full(a.size(), x, dtype=a.dtype, device=a.device)

>>> torch.jit.trace(fn1, torch.tensor(1.))(torch.tensor(2.))
tensor([1, 1])

>>> def fn2(x):
        return torch.empty_like(a).fill_(x)

>>> torch.jit.trace(fn1, torch.tensor(1.))(torch.tensor(2.))
tensor([2, 2])

[fritzo]

Interesting, it would be good to have this example in tests/infer/test_jit.py

[neerajprad]

Many thanks for updating the codebase! Looks great.

[neerajprad]

Will merge when tests pass.

[fehiepsi]

(I'll let @fehiepsi handle pyro.contrib.gp)

Yes yes, I'll do it. :D