Error while embedding: could not convert string to float: 'ng'

Question

I am working on Pre trained word vectors using GloVe method. Data contains vectors on Wikipedia data. While embedding data i am getting error stating that could not convert string to float: 'ng'

I tried going through data but there i was not able to find symbol 'ng'

# load embedding as a dict
def load_embedding(filename):
    # load embedding into memory, skip first line
    file = open(filename,'r', errors = 'ignore')
    # create a map of words to vectors
    embedding = dict()
    for line in file:
        parts = line.split()
        # key is string word, value is numpy array for vector
        embedding[parts[0]] = np.array(parts[1:], dtype='float32')
    file.close()
    return embedding

Here is the error report. Please guide me further.

runfile('C:/Users/AKSHAY/Desktop/NLP/Pre-trained GloVe.py', wdir='C:/Users/AKSHAY/Desktop/NLP')
C:\Users\AKSHAY\AppData\Local\conda\conda\envs\py355\lib\site-packages\h5py\__init__.py:36: FutureWarning: Conversion of the second argument of issubdtype from `float` to `np.floating` is deprecated. In future, it will be treated as `np.float64 == np.dtype(float).type`.
  from ._conv import register_converters as _register_converters
Using TensorFlow backend.
Traceback (most recent call last):

  File "<ipython-input-1-d91aa5ebf9f8>", line 1, in <module>
    runfile('C:/Users/AKSHAY/Desktop/NLP/Pre-trained GloVe.py', wdir='C:/Users/AKSHAY/Desktop/NLP')

  File "C:\Users\AKSHAY\AppData\Local\conda\conda\envs\py355\lib\site-packages\spyder\utils\site\sitecustomize.py", line 705, in runfile
    execfile(filename, namespace)

  File "C:\Users\AKSHAY\AppData\Local\conda\conda\envs\py355\lib\site-packages\spyder\utils\site\sitecustomize.py", line 102, in execfile
    exec(compile(f.read(), filename, 'exec'), namespace)

  File "C:/Users/AKSHAY/Desktop/NLP/Pre-trained GloVe.py", line 123, in <module>
    raw_embedding = load_embedding('glove.6B.50d.txt')

  File "C:/Users/AKSHAY/Desktop/NLP/Pre-trained GloVe.py", line 67, in load_embedding
    embedding[parts[0]] = np.array(parts[1:], dtype='float32')

ValueError: could not convert string to float: 'ng'

Answer 1

ValueError: could not convert string to float: 'ng'

For addressing the problem above, add encoding='utf8' to the function as follows:

file = open(filename,'r', errors = 'ignore', encoding='utf8')

Answer 2

Looks like 'ng' is a word (token) in your file that you are trying to get a word vector for. Glove pre-trained vectors probably do not have a vector for 'ng' which is causing the error. So, you need to check if the word has a vector in the Glove embeddings. See the section labeled 'Create a weight matrix for words in training docs' in this post for an example of how to do this - Text Classification Using CNN, LSTM and Pre-trained Glove Word Embeddings: Part-3

Answer 3

This seems to work fine:

embedding_model = {}
f = open(r'dataset/glove.840B.300d.txt', encoding="utf8", "r")
for line in f:
    values = line.split()
    word = ''.join(values[:-300])
    coefs = np.asarray(values[-300:], dtype='float32')
    embedding_model[word] = coefs
f.close()

Answer 4

You can do like this when using file glove.840B.300d.txt:

embedding_dict = {}
with open('glove.840B.300d.txt','r') as f:
    for line in f:
        values = line.split()
        word = ''.join(values[:-300])
        vectors = np.asarray(values[-300:], dtype='float32')
        embedding_dict[word] = vectors
f.close()