Philipp Dowling

Thank @arunchaganty! I had noticed that too and changed it. I was able to run `make all` after changing my gurobi license path and manually running `make install-requirements` (which I...

Another follow up - scoring also still fails in the out-of-the-box setting. Training runs for a while, I see lots of output, and then this: ``` Best objective 2.723835661325e+01, best...

Did you try anything? Do you have some code to share and your error message? In principle, AttentionSeq2Seq takes the same input as any other Seq2Seq model. See some usage...

You could probably look at the AttentionDecoderCell, and use that outside of the usual AttentionSeq2Seq wrapper (still wrapping it in a RecurrentContainer though). That cell should basically be able to...

Can you reformat the code to show what you mean? Is the problem that the attention is calculated using the encoder output sequence rather than the encoder input? I'm not...

Just to follow up: @farizrahman4u do you think there might be a problem here?

We need the ability to treat different channels, which 1-D convolutions can't do. We apply the 2-D convolution in a way that it is actually one dimensional, that's why there...