bug fixes and avoiding deprecation warnings

[michaelpradel]

Running Train.py with Python 3.6 and Pytorch 1.0.1 results in several deprecation warnings due to API changes, and eventually crashes with a TypeError (see below). This commit updates the code to Python 3.6 and Pytorch 1.0.1. Afterwards, it runs without any warnings or other problems.

Warnings:

PointerNet/PointerNet.py:105: UserWarning: nn.init.uniform is now deprecated in favor of nn.init.uniform_. nn.init.uniform(self.V, -1, 1) PointerNet/PointerNet.py:290: UserWarning: nn.init.uniform is now deprecated in favor of nn.init.uniform_. nn.init.uniform(self.decoder_input0, -1, 1)

Trace of crash:

... File "PointerNet/PointerNet.py", line 312, in forward decoder_hidden0 = (torch.cat(encoder_hidden[0][-2:], dim=-1), TypeError: cat(): argument 'tensors' (position 1) must be tuple of Tensors, not Tensor