Bruce Ho

icon location Working as a data scientist. Interested in advanced algorithms like hierarchical bayesian, mcmc, hmm, etc. as well as all deep learning architectures.

Results 33 comments of Bruce Ho

OOV during decoding

The vocab is fixed size throughout. I am looking into this issue in depth. The way OOV is predicted during decoding is really only meaning for during training, where the...

When pointer_gen=True, why did you change the decoder inputs using encoder oovs?

I am looking at the same issue. I believe it is because the word embedding is limited to the vocab size. These embeddings are updated during training, therefore the words...

install tensorflow-gpu==1.5.0 Issue

I am lost why you are talking about tensorflow. This code is written in chainer. did I miss something?

factor initialization

found this https://stackoverflow.com/questions/38426349/how-to-create-random-orthonormal-matrix-in-python-numpy

factor initialization

Having determined that the initialization aims to generate orthonormal random vectors, I see no basis in that requirement at all from the original paper. What would you want to impose...

RuntimeError: dim() called on undefined Tensor

is there any resolution to this? facing the same problem. it would really help if there is some documentation on which version of pytorch the code does work with.

RuntimeError: dim() called on undefined Tensor

I am using torch 0.4.0, and its failing. Maybe 0.3.0? if you have working code, please provide that version.

RuntimeError: dim() called on undefined Tensor

finally! 0.1.12 worked. Thank you very much.

what did you intend to do with drug_name.split("+")]

I used the re split with `plus_pattern = r'(?

what did you intend to do with drug_name.split("+")]

yes, I do understand the intent for the '+' in the code. my point is the code for splitting by '+' will fail for the dataset, because of drug names...