Python lists have no shape attribute about fightin-words HOT 4 CLOSED

Thanks for the bug!

Note: If you're pulling this from Pip or Anaconda... give it (and me) a while to update there. It's been a while since I looked at Python and I need to remember how to re-deploy :)

from fightin-words.

Thanks for the quick response!

from fightin-words.

Just one question if you have some time to help:

I managed to run everything successfully and am getting an output. As you may be able to infer from my submission above, I am trying to perform a calculation with an informative prior. However, even in the example provided, I am struggling to understand how the output is coherent according to the paper. I must be either misinterpreting the output, or the paper itself.
What I mean by the above is - when I run the provided example, the output is:

[('brown', -0.11159788992602564), ('donkey', -0.11159788992602564), ('fox', -0.11159788992602564), ('lazier', -0.11159788992602564), ('purple', -0.11159788992602564), ('quick', -0.11159788992602564), ('jumps', 0.04795579532980462), ('lazy', 0.04795579532980462), ('over', 0.04795579532980462), ('pig', 0.04795579532980462), ('the', 0.22827369129510436)]

Given that a comparison is taking place between l1 and l2, how can it be that "brown" and "purple" have the same scores, knowing that they belong to different classes and each word shows up exactly once in its respective class? My expectation is that "brown", showing up only on l1, would have a positive score that would be the absolute value of that of "purple".

I fear I might be grossly misinterpreting the output.

from fightin-words.

I had a discussion with @jmhessel over at his version of the implementation and managed to solve the issue above, but only when using his implementation. I can't figure out still why I am getting such an abnormal output when running your code and I do believe it is a mistake on my end. Once again, thanks a lot for your support and feedback!

from fightin-words.