Merge pull request #254 from huggingface/python_2

Adding OpenAI GPT and Transformer-XL models, compatibility with Python 2

pytorch_pretrained_bert/tokenization.py

@@ -14,14 +14,13 @@
 # limitations under the License.
 """Tokenization classes."""
 
-from __future__ import absolute_import
-from __future__ import division
-from __future__ import print_function
+from __future__ import absolute_import, division, print_function, unicode_literals
 
 import collections
-import unicodedata
-import os
 import logging
+import os
+import unicodedata
+from io import open
 
 from .file_utils import cached_path
 
@@ -117,26 +116,26 @@ class BertTokenizer(object):
         return tokens
 
     @classmethod
-    def from_pretrained(cls, pretrained_model_name, cache_dir=None, *inputs, **kwargs):
+    def from_pretrained(cls, pretrained_model_name_or_path, cache_dir=None, *inputs, **kwargs):
         """
         Instantiate a PreTrainedBertModel from a pre-trained model file.
         Download and cache the pre-trained model file if needed.
         """
-        if pretrained_model_name in PRETRAINED_VOCAB_ARCHIVE_MAP:
-            vocab_file = PRETRAINED_VOCAB_ARCHIVE_MAP[pretrained_model_name]
+        if pretrained_model_name_or_path in PRETRAINED_VOCAB_ARCHIVE_MAP:
+            vocab_file = PRETRAINED_VOCAB_ARCHIVE_MAP[pretrained_model_name_or_path]
         else:
-            vocab_file = pretrained_model_name
+            vocab_file = pretrained_model_name_or_path
         if os.path.isdir(vocab_file):
             vocab_file = os.path.join(vocab_file, VOCAB_NAME)
         # redirect to the cache, if necessary
         try:
             resolved_vocab_file = cached_path(vocab_file, cache_dir=cache_dir)
-        except FileNotFoundError:
+        except EnvironmentError:
             logger.error(
                 "Model name '{}' was not found in model name list ({}). "
                 "We assumed '{}' was a path or url but couldn't find any file "
                 "associated to this path or url.".format(
-                    pretrained_model_name,
+                    pretrained_model_name_or_path,
                     ', '.join(PRETRAINED_VOCAB_ARCHIVE_MAP.keys()),
                     vocab_file))
             return None
@@ -145,10 +144,10 @@ class BertTokenizer(object):
         else:
             logger.info("loading vocabulary file {} from cache at {}".format(
                 vocab_file, resolved_vocab_file))
-        if pretrained_model_name in PRETRAINED_VOCAB_POSITIONAL_EMBEDDINGS_SIZE_MAP:
+        if pretrained_model_name_or_path in PRETRAINED_VOCAB_POSITIONAL_EMBEDDINGS_SIZE_MAP:
             # if we're using a pretrained model, ensure the tokenizer wont index sequences longer
             # than the number of positional embeddings
-            max_len = PRETRAINED_VOCAB_POSITIONAL_EMBEDDINGS_SIZE_MAP[pretrained_model_name]
+            max_len = PRETRAINED_VOCAB_POSITIONAL_EMBEDDINGS_SIZE_MAP[pretrained_model_name_or_path]
             kwargs['max_len'] = min(kwargs.get('max_len', int(1e12)), max_len)
         # Instantiate tokenizer.
         tokenizer = cls(resolved_vocab_file, *inputs, **kwargs)

setup.py

@@ -33,12 +33,13 @@ To create the package for pypi.
 7. Copy the release notes from RELEASE.md to the tag in github once everything is looking hunky-dory.
 
 """
+from io import open
 from setuptools import find_packages, setup
 
 setup(
     name="pytorch_pretrained_bert",
-    version="0.4.0",
-    author="Thomas Wolf, Victor Sanh, Tim Rault, Google AI Language Team Authors",
+    version="0.5.0",
+    author="Thomas Wolf, Victor Sanh, Tim Rault, Google AI Language Team Authors, Open AI team Authors",
     author_email="thomas@huggingface.co",
     description="PyTorch version of Google AI BERT model with script to load Google pre-trained models",
     long_description=open("README.md", "r", encoding='utf-8').read(),
@@ -55,10 +56,10 @@ setup(
                       'tqdm'],
     entry_points={
       'console_scripts': [
-        "pytorch_pretrained_bert=pytorch_pretrained_bert.__main__:main"
+        "pytorch_pretrained_bert=pytorch_pretrained_bert.__main__:main",
       ]
     },
-    python_requires='>=3.5.0',
+    # python_requires='>=3.5.0',
     tests_require=['pytest'],
     classifiers=[
           'Intended Audience :: Science/Research',

tests/modeling_test.py

@@ -114,6 +114,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_model(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertModel(config=config)
+            model.eval()
             all_encoder_layers, pooled_output = model(input_ids, token_type_ids, input_mask)
             outputs = {
                 "sequence_output": all_encoder_layers[-1],
@@ -134,6 +135,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_masked_lm(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForMaskedLM(config=config)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, token_labels)
             prediction_scores = model(input_ids, token_type_ids, input_mask)
             outputs = {
@@ -149,6 +151,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_next_sequence_prediction(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForNextSentencePrediction(config=config)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, sequence_labels)
             seq_relationship_score = model(input_ids, token_type_ids, input_mask)
             outputs = {
@@ -165,6 +168,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_pretraining(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForPreTraining(config=config)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, token_labels, sequence_labels)
             prediction_scores, seq_relationship_score = model(input_ids, token_type_ids, input_mask)
             outputs = {
@@ -185,6 +189,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_question_answering(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForQuestionAnswering(config=config)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, sequence_labels, sequence_labels)
             start_logits, end_logits = model(input_ids, token_type_ids, input_mask)
             outputs = {
@@ -205,6 +210,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_sequence_classification(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForSequenceClassification(config=config, num_labels=self.num_labels)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, sequence_labels)
             logits = model(input_ids, token_type_ids, input_mask)
             outputs = {
@@ -221,6 +227,7 @@ class BertModelTest(unittest.TestCase):
 
         def create_bert_for_token_classification(self, config, input_ids, token_type_ids, input_mask, sequence_labels, token_labels):
             model = BertForTokenClassification(config=config, num_labels=self.num_labels)
+            model.eval()
             loss = model(input_ids, token_type_ids, input_mask, token_labels)
             logits = model(input_ids, token_type_ids, input_mask)
             outputs = {

tests/tokenization_test.py

@@ -12,15 +12,17 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from __future__ import absolute_import
-from __future__ import division
-from __future__ import print_function
+from __future__ import absolute_import, division, print_function, unicode_literals
 
 import os
 import unittest
+from io import open
 
-from pytorch_pretrained_bert.tokenization import (BertTokenizer, BasicTokenizer, WordpieceTokenizer,
-                                                  _is_whitespace, _is_control, _is_punctuation)
+from pytorch_pretrained_bert.tokenization import (BasicTokenizer,
+                                                  BertTokenizer,
+                                                  WordpieceTokenizer,
+                                                  _is_control, _is_punctuation,
+                                                  _is_whitespace)
 
 
 class TokenizationTest(unittest.TestCase):
@@ -30,7 +32,7 @@ class TokenizationTest(unittest.TestCase):
             "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn",
             "##ing", ","
         ]
-        with open("/tmp/bert_tokenizer_test.txt", "w") as vocab_writer:
+        with open("/tmp/bert_tokenizer_test.txt", "w", encoding='utf-8') as vocab_writer:
             vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
 
             vocab_file = vocab_writer.name
@@ -49,7 +51,7 @@ class TokenizationTest(unittest.TestCase):
             "[UNK]", "[CLS]", "[SEP]", "want", "##want", "##ed", "wa", "un", "runn",
             "##ing", ","
         ]
-        with open("/tmp/bert_tokenizer_test.txt", "w") as vocab_writer:
+        with open("/tmp/bert_tokenizer_test.txt", "w", encoding='utf-8') as vocab_writer:
             vocab_writer.write("".join([x + "\n" for x in vocab_tokens]))
             vocab_file = vocab_writer.name