spacy_bytes_load.py

def load_data(filename):
  
    traina =  cPickle.load( open( filename, "rb" ) )

    j = 0
    i = 0
    temp = None
    pairs = []
    
    for r in traina:
        d = spacy.tokens.doc.Doc(nlp.vocab)
        d.from_bytes(r)
        e = spacy.tokens.doc.Doc(nlp.vocab, words=[unicode(w) for w in d if not w.is_stop])

        if i % 2 == 1:
            pairs.append((e, temp))
            j += 1
        else:
            temp = e

        if i % 100 == 0:
            print(i)

        i += 1
       
     return pairs