model = Doc2Vec(min_count=1, window=10, size=200, seed=SEED, sample=1e-4,
                alpha=0.025, negative=5, workers=24)
model.build_vocab([x for x in tqdm(train_documents)])
train_documents  = shuffle(train_documents)
model.train(train_documents,total_examples=len(train_documents), epochs=30)

def vector_for_learning(model, input_docs):
    sents = input_docs
    targets, feature_vectors = zip(*[(doc.tags[0], model.infer_vector(doc.words, steps=20)) for doc in sents])
    return targets, feature_vectors
RAW Paste Data