Vivek Narayanan

That's correct, "oh my god" has a much greater negative weight than the positive weight of "love" in the training set. That said this model works better on longer sequences...

While this is a feature that would be great to have, the implementation might be tricky and I'm not sure if I follow the discussion completely. I have a few...

There is no `/web/text/` directory, it is a dynamic URL route which is handled by the function `evaldata()` in server.py. How are you running the server?

The nohup part is just to make it run in the background after you quit the terminal or ssh session. Try without nohup ( `python run.py` ) and you'll get...

Did you install all the dependencies using pip? Setting `app.debug = True` in the server.py file might be useful here.

The HOST variable in the config file is just for rendering HTML. Just noticed that your user agent says "Windows", are you using Windows to run the server? I'm not...

Oh, in that case edit this line in run.py `http_server = WSGIServer(('', config.PORT), app)` and change it to `http_server = WSGIServer(('your-host-name', config.PORT), app)`