Exemple #1
0
def getData_new_method(USE_BPE):
    if USE_BPE == False:
        g_tok = get_tokenizer('spacy', language='de')
        e_tok = get_tokenizer('spacy', language='en')
        return getData2(g_tok, e_tok, USE_BPE)
    else:
        if config.BPE_FROM_PICKLE:
            pkl_file = open('BPE/data.pkl', 'rb')
            data1 = pickle.load(pkl_file)
            sp_gec = data1["sp_gec_orig"]
        else:
            sp_gec = load_sp_model(config.BPE_PATH)
        return getData2(sp_gec, sp_gec, USE_BPE)
Exemple #2
0
def getData_new_method():
    g_tok = get_tokenizer('spacy', language='en')
    e_tok = get_tokenizer('spacy', language='de')
    return getData2(g_tok, e_tok)