aeb516494daa207720b428072ab49eeea7bfff75,scripts/text_generation/sequence_sampling.py,,,#,33

Before Change

                                             "We load a LSTM model that is pre-trained on "
                                             "WikiText as our encoder.")
group = parser.add_mutually_exclusive_group(required=True)
group.add_argument("--use-sampling", action="store_true",
                   help="Use sampling instead of beam search.")
group.add_argument("--use-beam-search", action="store_true",
                   help="Use beam search instead of random sampling.")
parser.add_argument("--lm_model", type=str, default="awd_lstm_lm_1150",
                    help="type of the pre-trained model to load, can be "standard_lstm_lm_200", "

After Change

// beam search sampler options
subparsers = parser.add_subparsers(help="Sequence generation methods.",
                                   dest="command")
subparsers.required = True
beam_search_parser = subparsers.add_parser("beam-search", help="Use beam search for decoding.")
beam_search_parser.add_argument("--alpha", type=float, default=0.0,
                                help="Alpha in the length penalty term.")
beam_search_parser.add_argument("--k", type=int, default=5, help="K in the length penalty term.")

// random sampler options
random_sample_parser = subparsers.add_parser("random-sample",
                                             help="Use random sampling for decoding.")
random_sample_parser.add_argument("--temperature", type=float, default=1.0,
                                  help="Softmax temperature used in sampling.")
random_sample_parser.add_argument("--use-top-k", type=int, required=False,
                                  help="Sample only from the top-k candidates.")

// shared options
for p in [beam_search_parser, random_sample_parser]:
    p.add_argument("--gpu", type=int, default=0,
                   help="id of the gpu to use. Set it to empty means to use cpu.")
    p.add_argument("--lm-model", type=str, default="awd_lstm_lm_1150",
                   help="type of the pre-trained model to load, can be "standard_lstm_lm_200", "
                        ""standard_lstm_lm_650", "standard_lstm_lm_1500", "
                        ""awd_lstm_lm_1150", etc.")
    p.add_argument("--max-length", type=int, default=20, help="Maximum sentence length.")
    p.add_argument("--print-num", type=int, default=3, help="Number of sentences to display.")
    p.add_argument("--bos", type=str, default="I think this works")
    p.add_argument("--beam-size", type=int, default=5,
                   help="Beam size in the beam search sampler.")

args = parser.parse_args()

print(args)
if args.gpu is not None and args.gpu < mx.context.num_gpus():

Instances