corrected dependencies

README.md

@@ -19,17 +19,17 @@ On a 32-core computer, 25 epochs of AdaGrad run in 3.8 hours on Wikipedia cooccu
 
 Note that our code takes as input an upper-triangular, zero-indexed cooccurrence matrix rather than the full, one-indexed cooccurrence matrix used by the original GloVe code. To convert to our (more disk-memory efficient) version you can use the method <tt>reformat_coocfile</tt> in <tt>solvers.py</tt>. We also allow direct, parallel computation of the vocab and cooccurrence files.
 
-Dependencies: NumPy, [SharedArray](https://pypi.org/project/SharedArray/)
+Dependencies: numpy, numba, [SharedArray](https://pypi.org/project/SharedArray/)
 
-Optional: h5py, mpi4py*, scikit-learn
+Optional: h5py, mpi4py*, scipy, scikit-learn
 
 \* required for parallelism; [MPI](http://www.mpich.org/downloads/) can be easily installed on Linux, Mac, and Windows Subsystem for Linux
 
 # DisC embeddings
 
 Scripts to recreate the results in the paper are provided in the directory <tt>scripts-AKSV2018</tt>. 1600-dimensional GloVe embeddings trained on the Amazon Product Corpus [3] are provided [here](http://nlp.cs.princeton.edu/DisC/amazon_glove1600.txt.bz2).
 
-Dependencies: NLTK, NumPy, SciPy, scikit-learn
+Dependencies: nltk, numpy, scipy, scikit-learn
 
 Optional: tensorflow    
 

solvers.py

@@ -8,10 +8,9 @@
 from collections import deque
 from operator import itemgetter
 from tempfile import NamedTemporaryFile as NTF
-import h5py
+import SharedArray as sa
 import numpy as np
 from numba import jit
-from sklearn.linear_model import LinearRegression as LR
 from text_embedding.documents import *
 
 
@@ -254,8 +253,6 @@ def __enter__(self):
 
     def __exit__(self, *args):
 
-        import SharedArray as sa
-
         for array in self._shared:
             try:
                 sa.delete(array)
@@ -264,8 +261,6 @@ def __exit__(self, *args):
 
     def create(self, array=None, dtype=None):
 
-        import SharedArray as sa
-
         comm, rank = self._comm, self._rank
 
         if rank:
@@ -410,6 +405,8 @@ def save(self, fid):
             None
         '''
 
+        import h5py
+
         if not self._rank:
             f = h5py.File(fid)
             for name, param in zip(self._pnames, self._params[:self._numpar]):
@@ -832,6 +829,7 @@ def align_params(params, srcvocab, tgtvocab, mean_fill=True):
 def induce_embeddings(srcvocab, srccooc, srcvecs, tgtvocab, tgtcooc, comm=None):
 
     from scipy import sparse as sp
+    from sklearn.linear_model import LinearRegression as LR
 
     rank, size = ranksize(comm)
     Vsrc, d = srcvecs.shape