a886baab6b48d976f0b30addde5e588282de072f,util/select_pairs_for_ab.py,,main,#,60

Before Change

    with codecs.open(args.out_file, "wb", "UTF-8)") as fh:
        writer = csv.writer(fh, delimiter=b",", lineterminator=b"\n")
        writer.writerow(("context", "text_a", "text_b", "origin_a", "origin_b"))
        for context, inst1, inst2 in data:
            if random.random() >= 0.5:
                writer.writerow((context, inst2, inst1, args.file2, args.file1))
            else:

After Change

    else:
        data2 = parse_sgm_file(args.file2)
        data = create_real_questions(contexts, data1, data2)
        header = ("context", "text_a", "text_b", "origin_a", "origin_b")

    if len(data) < args.num_samples:
        print >> sys.stderr, "Not enough samples, generating only %d." % len(data)

    random.shuffle(data)
    data = data[:args.num_samples]

    with codecs.open(args.out_file, "wb", "UTF-8)") as fh:
        writer = csv.writer(fh, delimiter=b",", lineterminator=b"\n")
        writer.writerow(header)
        for context, inst1, inst2 in data:
            if args.test_questions:
                if random.random() >= 0.5:
                    writer.writerow((context, inst2, inst1, args.file1 + "-distorted", args.file1,

Instances