teucer
teucer
Having `include_intercept` in the constructor would be really helpful.
@rodgzilla have seen it, thx. It seems that `self.clf_dropout` is not used used anywhere. Do we need it at all?
Had the same issue with imdb sentiment analysis. Would appreciate some pointers here...
@rodgzilla ok will do that. What about increasing the dropout probability in the classification head? Would it help to increase it?
@rodgzilla Thank you a lot for the explanation. It makes a lot of sense! Out of curiosity, why all the values cannot be used?
What I don't understand is why we need to start from `n_vocab + n_special` not e.g. from 0. Any explanations?
That might come handy for some other use cases: the code blocks could be executed and their output put back into markdown file.
I assume that the file is read into memory. Unfortunately the linter and the formatter (at least the python ones) are using physical file locations. This is why having a...
Some further thoughts: 1. Don't know if it would help to think about architecture now or wait for the "final" solution before refactoring. There are a pros and cons with...