keitakurita
keitakurita
Previously, models with multiple inputs would fail during training due with shuffle=True due to a bug in the shuffler in the multiple input helpers. This quick fix solves this bug.
Beautiful soup is a lot slower than lxml, and lxml should suffice for the task of extracting info from tweets
Making the parser run in parallel might make the crawler run faster.
Release an error message that more explicitly tells the user the url that failed to connect
