PEWOLF

Based on my understanding of [this](https://stackoverflow.com/questions/37587622/tf-nn-in-top-k-targets-out-of-range) issue in stackoverflow, I modified the code so that the the 'cell's passed as argument to _HANClassifierModel_ in _sentence_cell_ and _word_cell_ become functions returning...

I don't have the logs anymore, but I remember it was more or less the same speed that you have. Some implementations do not properly leverage GPUs power. I advice...

The section about word embedding in the paper's paragraph 2.2: > Note that we directly use word embeddings. For a more complete model we could use a GRU to get...

I found the problem with interpolation: interpolating between a sentence and itself gave different sentences because of the random sampling. I set top_k to 1 and top_p to 0.0 and...

Thanks a lot for this article ! Thanks to you, I recoded parts of my model without ```tf.map_fn( lambda: ....)``` and got an acceleration of ~x100 ! I basically repeated...