Use Amazon S3 Transfer Acceleration

scripts/datasets/language_modeling/prepare_lm.py

@@ -50,9 +50,9 @@
     # The original address of Google One Billion Word dataset is
     # http://www.statmt.org/lm-benchmark/1-billion-word-language-modeling-benchmark-r13output.tar.gz
     # We uploaded the file to S3 to accelerate the speed
-    'gbw': 'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/datasets/language_modeling/1-billion-word-language-modeling-benchmark-r13output.tar.gz',
+    'gbw': 'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/datasets/language_modeling/1-billion-word-language-modeling-benchmark-r13output.tar.gz',
     # The data is obtained from https://raw.githubusercontent.com/rafaljozefowicz/lm/master/1b_word_vocab.txt
-    'gbw_vocab': 'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/datasets/language_modeling/1b_word_vocab.txt'
+    'gbw_vocab': 'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/datasets/language_modeling/1b_word_vocab.txt'
 }
 
 

scripts/datasets/machine_translation/prepare_wmt.py

@@ -235,7 +235,7 @@
     # For the CWMT dataset, you can also download them from the official location: http://nlp.nju.edu.cn/cwmt-wmt/
     # Currently, this version is processed via https://gist.github.com/sxjscience/54bedd68ce3fb69b3b1b264377efb5a5
     'cwmt': {
-        'url': 'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt.tar.gz',
+        'url': 'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt.tar.gz',
         'zh-en': {
             'en': 'cwmt/cwmt-zh-en.en',
             'zh': 'cwmt/cwmt-zh-en.zh'

scripts/datasets/pretrain_corpus/prepare_bookcorpus.py

@@ -34,7 +34,7 @@
 
 _URLS = {
     'gutenberg':
-        'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/pretrain_corpus/Gutenberg.zip',
+        'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/pretrain_corpus/Gutenberg.zip',
 }
 
 

scripts/datasets/url_checksums/book_corpus.txt

@@ -1 +1 @@
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/pretrain_corpus/Gutenberg.zip 91e842dc3671ed5a917b7ff6a60f5f87397780e2 461506225
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/pretrain_corpus/Gutenberg.zip 91e842dc3671ed5a917b7ff6a60f5f87397780e2 461506225

scripts/datasets/url_checksums/language_model.txt

@@ -2,5 +2,5 @@ https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-v1.zip 3c914d1
 https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-v1.zip 0aec09a7537b58d4bb65362fee27650eeaba625a 190229076
 http://mattmahoney.net/dc/enwik8.zip d856b1ccd937c51aeb9c342e47666fb8c38e7e72 36445475
 http://mattmahoney.net/dc/text8.zip 6c70299b93b7e1f927b42cd8f6ac1a31547c7a2e 31344016
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/datasets/language_modeling/1-billion-word-language-modeling-benchmark-r13output.tar.gz 4df859766482e12264a5a9d9fb7f0e276020447d 1792209805
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/datasets/language_modeling/1b_word_vocab.txt aa2322a3da82ef628011336c9b5c6059e4f56c3f 9507106
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/datasets/language_modeling/1-billion-word-language-modeling-benchmark-r13output.tar.gz 4df859766482e12264a5a9d9fb7f0e276020447d 1792209805
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/datasets/language_modeling/1b_word_vocab.txt aa2322a3da82ef628011336c9b5c6059e4f56c3f 9507106

scripts/datasets/url_checksums/wmt.txt

@@ -34,16 +34,16 @@ https://stuncorpusprod.blob.core.windows.net/corpusfiles/UNv1.0.en-ru.tar.gz.01
 https://stuncorpusprod.blob.core.windows.net/corpusfiles/UNv1.0.en-ru.tar.gz.02 bf6b18a33c8cafa6889fd463fa8a2850d8877d35 306221588
 https://stuncorpusprod.blob.core.windows.net/corpusfiles/UNv1.0.en-zh.tar.gz.00 1bec5f10297512183e483fdd4984d207700657d1 1073741824
 https://stuncorpusprod.blob.core.windows.net/corpusfiles/UNv1.0.en-zh.tar.gz.01 15df2968bc69ef7662cf3029282bbb62cbf107b1 312943879
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt/parallel/casia2015.zip b432394685e4c53797e1ac86851f8a013aef27a2 98159063
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt/parallel/casict2011.zip 769a9a86c24e9507dbf520b950b9026120cb041e 166957775
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt/parallel/datum2015.zip 6d94cc8d296dd4268ed0a10fa3a419267280363e 100118018
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt/parallel/datum2017.zip 480fa06760b2dbe7c9a9bd7c3fd5e5b22b860a45 37389573
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt/parallel/neu2017.zip 532b56ba62f6cffccdc85f4316468873ca739bd1 148681171
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt/parallel/casia2015.zip b432394685e4c53797e1ac86851f8a013aef27a2 98159063
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt/parallel/casict2011.zip 769a9a86c24e9507dbf520b950b9026120cb041e 166957775
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt/parallel/datum2015.zip 6d94cc8d296dd4268ed0a10fa3a419267280363e 100118018
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt/parallel/datum2017.zip 480fa06760b2dbe7c9a9bd7c3fd5e5b22b860a45 37389573
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt/parallel/neu2017.zip 532b56ba62f6cffccdc85f4316468873ca739bd1 148681171
 http://data.statmt.org/wmt17/translation-task/rapid2016.tgz 8b173ce0bc77f2a1a57c8134143e3b5ae228a6e2 163416042
 https://s3-eu-west-1.amazonaws.com/tilde-model/rapid2019.de-en.zip aafe431338abb98fc20951b2d6011223a1b91311 111888392
 http://data.statmt.org/wmt19/translation-task/dev.tgz 451ce2cae815c8392212ccb3f54f5dcddb9b2b9e 38654961
 http://data.statmt.org/wmt19/translation-task/test.tgz ce02a36fb2cd41abfa19d36eb8c8d50241ed3346 3533424
-https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/wmt/cwmt.tar.gz 88c2f4295169e9f0a9834bf8bff87e3fd4c04055 709032378
+https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/wmt/cwmt.tar.gz 88c2f4295169e9f0a9834bf8bff87e3fd4c04055 709032378
 http://data.statmt.org/news-crawl/de/news.2007.de.shuffled.deduped.gz 9d746b9df345f764e6e615119113c70e3fb0858c 90104365
 http://data.statmt.org/news-crawl/de/news.2008.de.shuffled.deduped.gz 185a24e8833844486aee16cb5decf9a64da1c101 308205291
 http://data.statmt.org/news-crawl/de/news.2009.de.shuffled.deduped.gz 9f7645fc6467de88f4205d94f483194838bad8ce 317590378

src/gluonnlp/base.py

@@ -57,7 +57,7 @@ def get_model_zoo_checksum_dir():
 def get_repo_url():
     """Return the base URL for Gluon dataset and model repository """
     # TODO(sxjscience) Revise later by calling gluon.utils._get_repo_url
-    default_repo = 'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/'
+    default_repo = 'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/'
     repo_url = os.environ.get('MXNET_GLUON_REPO', default_repo)
     if repo_url[-1] != '/':
         repo_url = repo_url + '/'

src/gluonnlp/data/tokenizers.py

@@ -1259,7 +1259,7 @@ class SentencepieceTokenizer(BaseTokenizerWithVocab):
     Examples
     --------
     >>> from mxnet import gluon
-    >>> url = 'https://gluonnlp-numpy-data.s3-us-west-2.amazonaws.com/tokenizer_test_models/sentencepiece/test_ende-a9bee4.model'
+    >>> url = 'https://gluonnlp-numpy-data.s3-accelerate.amazonaws.com/tokenizer_test_models/sentencepiece/test_ende-a9bee4.model'
     >>> model_f = gluon.utils.download(url)
     -etc-
     >>> tokenizer = gluonnlp.data.SentencepieceTokenizer(model_f)