update hash (#332)

Author: eric-haibin-lin

gluonnlp/model/language_model.py

@@ -422,7 +422,7 @@ def standard_lstm_lm_1500(dataset_name=None, vocab=None, pretrained=False, ctx=c
         ('700b532dc96a29e39f45cb7dd632ce44e377a752', 'standard_lstm_lm_200_wikitext-2'),
         ('45d6df33f35715fb760ec8d18ed567016a897df7', 'awd_lstm_lm_1150_wikitext-2'),
         ('7894a046f8286db0d5d2ed672b60f4f52b4bc3aa', 'awd_lstm_lm_600_wikitext-2'),
-        ('92ed6f71c4b12f465b11d72e7271743d0c4c1a7a', 'big_rnn_lm_2048_512_gbw'),
+        ('6bb3e991eb4439fabfe26c129da2fe15a324e918', 'big_rnn_lm_2048_512_gbw'),
     ]})
 
 class BigRNN(Block):

scripts/language_model/language_model.rst

@@ -179,18 +179,18 @@ The dataset used for training the models is Google's 1 billion words dataset.
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
 | Graident clip   | 10.0                                                                                                                         |
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
-| Test perplexity | 43.72                                                                                                                        |
+| Test perplexity | 43.62                                                                                                                        |
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
-| Num epochs      | 42                                                                                                                           |
+| Num epochs      | 50                                                                                                                           |
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
 | Training logs   | `log <https://github.com/dmlc/web-data/blob/master/gluonnlp/logs/language_model/big_rnn_lm_2048_512_gbw.log>`__              |
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
 | Evaluation logs | `log <https://github.com/dmlc/web-data/blob/master/gluonnlp/logs/language_model/big_rnn_lm_2048_512_gbw-eval.log>`__         |
 +-----------------+------------------------------------------------------------------------------------------------------------------------------+
 
-[1] LSTM-2048-512 (Test PPL 43.72, based on `commit <https://github.com/eric-haibin-lin/gluon-nlp/tree/b3cfcc427fbceea2ce3c5fae63c61b93910326dc>`__)
+[1] LSTM-2048-512 (Test PPL 43.62)
 
 .. code-block:: console
 
-   $ python large_word_language_model.py --gpus 0,1,2,3 --epochs=42 --clip=10
-   $ python large_word_language_model.py --gpus 4 --eval-only --batch-size=1 --log-interval=999999
+   $ python large_word_language_model.py --gpus 0,1,2,3 --clip=10
+   $ python large_word_language_model.py --gpus 4 --eval-only --batch-size=1

scripts/language_model/large_word_language_model.py

@@ -83,7 +83,7 @@
                     help='list of gpus to run, e.g. 0 or 0,2,5. empty means using cpu.')
 parser.add_argument('--log-interval', type=int, default=1000,
                     help='report interval')
-parser.add_argument('--seed', type=int, default=1,
+parser.add_argument('--seed', type=int, default=0,
                     help='random seed')
 parser.add_argument('--lr', type=float, default=0.2,
                     help='initial learning rate')