(data_dir)
| 46 | |
| 47 | |
| 48 | def get_label2answer(data_dir): |
| 49 | import gzip |
| 50 | label2answer = dict() |
| 51 | with gzip.open(data_dir + |
| 52 | "/InsuranceQA.label2answer.token.encoded.gz") as fin: |
| 53 | for line in fin: |
| 54 | pair = line.decode().strip().split("\t") |
| 55 | idxs = pair[1].split(" ") |
| 56 | idxs = [int(idx.replace("idx_", "")) for idx in idxs] |
| 57 | label2answer[int(pair[0])] = idxs |
| 58 | return label2answer |
| 59 | |
| 60 | |
| 61 | pad_idx = 0 |