nikhgarg/embeddingdynamicstereotypes

embeddingdynamicstereotypes's Issues

Download all word embeddings

I am trying to re-run your code and have been struggling to find all the word embeddings you used during your experiments. Could you please let me know from where do I download them? Also, could you elaborate on exactly how do I run your code? The README does not help me a lot.

Thank You

How to normalize vectors by l2 norm?

download vectors from online sources and normalize by l2 norm (links in paper and below)

I am not very clear about this.

Any help would be appreciated.
Thank You.

Getting NaN after running changes_over_time.py

I have been trying to run the code since long and keep getting NaNs in the finalrun.csv which is generated after running changes_over_time.py. I have downloaded all embeddings twice (as I thought there was some error while downloading the vectors the first time I ran), normalized all vectors and then run changes_over_time.py.

Any help would be appreciated.

Thank You.

Input Data

Hello, thanks for the repository!

I have followed the instructions in the README file but cannot get the vectors as described here

Basically these files:

filenames_sgns = [folder + 'vectors_sgns{}.txt'.format(x) for x in range(1910, 2000, 10)]
filenames_svd = [folder + 'vectors_svd{}.txt'.format(x) for x in range(1910, 2000, 10)]
filenames_nyt = [folder + 'vectors{}-{}.txt'.format(x, x+5) for x in range(1987, 2000, 1)]
filenames_coha = [folder + 'vectorscoha{}-{}.txt'.format(x, x+20) for x in range(1910, 2000, 10)]

Can you please let us know how to generate them ? I have for example downloaded the sgns from here but it contains only -vocab.pkl and -w.npy but not any .txt files.

New York Times Annotated Corpus link returns 404

The link to the New York Times Annotated Corpus in README.md is wrong and returns 404.

The correct one is: https://catalog.ldc.upenn.edu/LDC2008T19