i am unable to use this code:
it is an open source machine learning model. I want to train my own decoder:
"Decoding: Generating the sentence that the conditioned vector had encoded"
i believe the instructions for the README are bad and the code might have some bugs, you can see in other people's forks that there are lots of bug fixes.
I want to be able to train my own skip-thought vector with the scifi genre from the bookcorpus dataset:
in their example model, they also train off the bookcorpus dataset, but only on the romance genre.
Here is a ticket with the exact same issue that I am having:
to consider this done, we need 2 things:
code that allows me to train on any arbitrary dataset.
a trained scifi bookcorpus model