Patch: v2.4.1

cc @julien-c

.circleci/deploy.sh

@@ -3,7 +3,7 @@ cd docs
 function deploy_doc(){
 	echo "Creating doc at commit $1 and pushing to folder $2"
 	git checkout $1
-	if [ ! -z "$2" ] 
+	if [ ! -z "$2" ]
 	then
 		if [ -d "$dir/$2" ]; then
 			echo "Directory" $2 "already exists"
@@ -17,7 +17,7 @@ function deploy_doc(){
 	fi
 }
 
-deploy_doc "master" 
+deploy_doc "master"
 deploy_doc "b33a385" v1.0.0
 deploy_doc "fe02e45" v1.1.0
 deploy_doc "89fd345" v1.2.0
@@ -25,3 +25,4 @@ deploy_doc "fc9faa8" v2.0.0
 deploy_doc "3ddce1d" v2.1.1
 deploy_doc "3616209" v2.2.0
 deploy_doc "d0f8b9a" v2.3.0
+deploy_doc "6664ea9" v2.4.0

README.md

@@ -60,7 +60,7 @@ Choose the right framework for every part of a model's lifetime
 | [Quick tour: Share your models ](#Quick-tour-of-model-sharing) | Upload and share your fine-tuned models with the community |
 | [Migrating from pytorch-transformers to transformers](#Migrating-from-pytorch-transformers-to-transformers) | Migrating your code from pytorch-transformers to transformers |
 | [Migrating from pytorch-pretrained-bert to pytorch-transformers](#Migrating-from-pytorch-pretrained-bert-to-transformers) | Migrating your code from pytorch-pretrained-bert to transformers |
-| [Documentation][(v2.3.0)](https://huggingface.co/transformers/v2.3.0)[(v2.2.0/v2.2.1/v2.2.2)](https://huggingface.co/transformers/v2.2.0) [(v2.1.1)](https://huggingface.co/transformers/v2.1.1) [(v2.0.0)](https://huggingface.co/transformers/v2.0.0) [(v1.2.0)](https://huggingface.co/transformers/v1.2.0) [(v1.1.0)](https://huggingface.co/transformers/v1.1.0) [(v1.0.0)](https://huggingface.co/transformers/v1.0.0) [(master)](https://huggingface.co/transformers) | Full API documentation and more |
+| [Documentation][(v2.4.0)](https://huggingface.co/transformers/v2.4.0)[(v2.3.0)](https://huggingface.co/transformers/v2.3.0)[(v2.2.0/v2.2.1/v2.2.2)](https://huggingface.co/transformers/v2.2.0) [(v2.1.1)](https://huggingface.co/transformers/v2.1.1) [(v2.0.0)](https://huggingface.co/transformers/v2.0.0) [(v1.2.0)](https://huggingface.co/transformers/v1.2.0) [(v1.1.0)](https://huggingface.co/transformers/v1.1.0) [(v1.0.0)](https://huggingface.co/transformers/v1.0.0) [(master)](https://huggingface.co/transformers) | Full API documentation and more |
 
 ## Installation
 

docs/source/conf.py

@@ -26,7 +26,7 @@ author = u'huggingface'
 # The short X.Y version
 version = u''
 # The full version, including alpha/beta/rc tags
-release = u'2.4.0'
+release = u'2.4.1'
 
 
 # -- General configuration ---------------------------------------------------

examples/run_generation.py

@@ -221,6 +221,7 @@ def main():
         top_k=args.k,
         top_p=args.p,
         repetition_penalty=args.repetition_penalty,
+        do_sample=True,
     )
 
     # Batch size == 1. to add more examples please use num_return_sequences > 1

setup.py

@@ -23,6 +23,8 @@ To create the package for pypi.
 
    twine upload dist/* -r pypitest
    (pypi suggest using twine as other methods upload files via plaintext.)
+   You may have to specify the repository url, use the following command then:
+   twine upload dist/* -r pypitest --repository-url=https://test.pypi.org/legacy/
 
    Check that you can install it in a virtualenv by running:
    pip install -i https://testpypi.python.org/pypi transformers
@@ -73,7 +75,7 @@ extras["dev"] = extras["testing"] + extras["quality"] + ["mecab-python3", "sciki
 
 setup(
     name="transformers",
-    version="2.4.0",
+    version="2.4.1",
     author="Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Google AI Language Team Authors, Open AI team Authors, Facebook AI Authors, Carnegie Mellon University Authors",
     author_email="thomas@huggingface.co",
     description="State-of-the-art Natural Language Processing for TensorFlow 2.0 and PyTorch",

src/transformers/__init__.py

@@ -2,7 +2,7 @@
 # There's no way to ignore "F401 '...' imported but unused" warnings in this
 # module, but to preserve other warnings. So, don't check this module at all.
 
-__version__ = "2.4.0"
+__version__ = "2.4.1"
 
 # Work around to update TensorFlow's absl.logging threshold which alters the
 # default Python logging output behavior when present.

src/transformers/configuration_flaubert.py

@@ -50,8 +50,8 @@ class FlaubertConfig(XLMConfig):
                 Probability to drop layers during training (Fan et al., Reducing Transformer Depth on Demand
                 with Structured Dropout. ICLR 2020)
             vocab_size (:obj:`int`, optional, defaults to 30145):
-                Vocabulary size of the XLM model. Defines the different tokens that
-                can be represented by the `inputs_ids` passed to the forward method of :class:`~transformers.XLMModel`.
+                Vocabulary size of the Flaubert model. Defines the different tokens that
+                can be represented by the `inputs_ids` passed to the forward method of :class:`~transformers.FlaubertModel`.
             emb_dim (:obj:`int`, optional, defaults to 2048):
                 Dimensionality of the encoder layers and the pooler layer.
             n_layer (:obj:`int`, optional, defaults to 12):

src/transformers/modeling_auto.py

@@ -129,8 +129,8 @@ ALL_PRETRAINED_MODEL_ARCHIVE_MAP = dict(
         ALBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
         CAMEMBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
         T5_PRETRAINED_MODEL_ARCHIVE_MAP,
-        XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_MAP,
         FLAUBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
+        XLM_ROBERTA_PRETRAINED_MODEL_ARCHIVE_MAP,
     ]
     for key, value, in pretrained_map.items()
 )
@@ -148,9 +148,9 @@ MODEL_MAPPING = OrderedDict(
         (GPT2Config, GPT2Model),
         (TransfoXLConfig, TransfoXLModel),
         (XLNetConfig, XLNetModel),
+        (FlaubertConfig, FlaubertModel),
         (XLMConfig, XLMModel),
         (CTRLConfig, CTRLModel),
-        (FlaubertConfig, FlaubertModel),
     ]
 )
 
@@ -167,9 +167,9 @@ MODEL_FOR_PRETRAINING_MAPPING = OrderedDict(
         (GPT2Config, GPT2LMHeadModel),
         (TransfoXLConfig, TransfoXLLMHeadModel),
         (XLNetConfig, XLNetLMHeadModel),
+        (FlaubertConfig, FlaubertWithLMHeadModel),
         (XLMConfig, XLMWithLMHeadModel),
         (CTRLConfig, CTRLLMHeadModel),
-        (FlaubertConfig, FlaubertWithLMHeadModel),
     ]
 )
 
@@ -186,9 +186,9 @@ MODEL_WITH_LM_HEAD_MAPPING = OrderedDict(
         (GPT2Config, GPT2LMHeadModel),
         (TransfoXLConfig, TransfoXLLMHeadModel),
         (XLNetConfig, XLNetLMHeadModel),
+        (FlaubertConfig, FlaubertWithLMHeadModel),
         (XLMConfig, XLMWithLMHeadModel),
         (CTRLConfig, CTRLLMHeadModel),
-        (FlaubertConfig, FlaubertWithLMHeadModel),
     ]
 )
 
@@ -201,8 +201,8 @@ MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING = OrderedDict(
         (RobertaConfig, RobertaForSequenceClassification),
         (BertConfig, BertForSequenceClassification),
         (XLNetConfig, XLNetForSequenceClassification),
-        (XLMConfig, XLMForSequenceClassification),
         (FlaubertConfig, FlaubertForSequenceClassification),
+        (XLMConfig, XLMForSequenceClassification),
     ]
 )
 
@@ -213,8 +213,8 @@ MODEL_FOR_QUESTION_ANSWERING_MAPPING = OrderedDict(
         (RobertaConfig, RobertaForQuestionAnswering),
         (BertConfig, BertForQuestionAnswering),
         (XLNetConfig, XLNetForQuestionAnswering),
-        (XLMConfig, XLMForQuestionAnswering),
         (FlaubertConfig, FlaubertForQuestionAnswering),
+        (XLMConfig, XLMForQuestionAnswering),
     ]
 )
 

src/transformers/tokenization_auto.py

@@ -25,6 +25,7 @@ from .configuration_auto import (
     CamembertConfig,
     CTRLConfig,
     DistilBertConfig,
+    FlaubertConfig,
     GPT2Config,
     OpenAIGPTConfig,
     RobertaConfig,
@@ -41,6 +42,7 @@ from .tokenization_bert_japanese import BertJapaneseTokenizer
 from .tokenization_camembert import CamembertTokenizer
 from .tokenization_ctrl import CTRLTokenizer
 from .tokenization_distilbert import DistilBertTokenizer
+from .tokenization_flaubert import FlaubertTokenizer
 from .tokenization_gpt2 import GPT2Tokenizer
 from .tokenization_openai import OpenAIGPTTokenizer
 from .tokenization_roberta import RobertaTokenizer
@@ -67,6 +69,7 @@ TOKENIZER_MAPPING = OrderedDict(
         (GPT2Config, GPT2Tokenizer),
         (TransfoXLConfig, TransfoXLTokenizer),
         (XLNetConfig, XLNetTokenizer),
+        (FlaubertConfig, FlaubertTokenizer),
         (XLMConfig, XLMTokenizer),
         (CTRLConfig, CTRLTokenizer),
     ]

tests/test_modeling_auto.py

@@ -39,6 +39,14 @@ if is_torch_available():
         BertForQuestionAnswering,
     )
     from transformers.modeling_bert import BERT_PRETRAINED_MODEL_ARCHIVE_MAP
+    from transformers.modeling_auto import (
+        MODEL_MAPPING,
+        MODEL_FOR_PRETRAINING_MAPPING,
+        MODEL_FOR_QUESTION_ANSWERING_MAPPING,
+        MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
+        MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
+        MODEL_WITH_LM_HEAD_MAPPING,
+    )
 
 
 @require_torch
@@ -127,3 +135,26 @@ class AutoModelTest(unittest.TestCase):
         self.assertIsInstance(model, RobertaForMaskedLM)
         self.assertEqual(model.num_parameters(), 14830)
         self.assertEqual(model.num_parameters(only_trainable=True), 14830)
+
+    def test_parents_and_children_in_mappings(self):
+        # Test that the children are placed before the parents in the mappings, as the `instanceof` will be triggered
+        # by the parents and will return the wrong configuration type when using auto models
+
+        mappings = (
+            MODEL_MAPPING,
+            MODEL_FOR_PRETRAINING_MAPPING,
+            MODEL_FOR_QUESTION_ANSWERING_MAPPING,
+            MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING,
+            MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING,
+            MODEL_WITH_LM_HEAD_MAPPING,
+        )
+
+        for mapping in mappings:
+            mapping = tuple(mapping.items())
+            for index, (child_config, child_model) in enumerate(mapping[1:]):
+                for parent_config, parent_model in mapping[: index + 1]:
+                    with self.subTest(
+                        msg="Testing if {} is child of {}".format(child_config.__name__, parent_config.__name__)
+                    ):
+                        self.assertFalse(issubclass(child_config, parent_config))
+                        self.assertFalse(issubclass(child_model, parent_model))

tests/test_tokenization_auto.py

@@ -25,6 +25,7 @@ from transformers import (
     GPT2Tokenizer,
     RobertaTokenizer,
 )
+from transformers.tokenization_auto import TOKENIZER_MAPPING
 
 from .utils import DUMMY_UNKWOWN_IDENTIFIER, SMALL_MODEL_IDENTIFIER, slow  # noqa: F401
 
@@ -70,3 +71,19 @@ class AutoTokenizerTest(unittest.TestCase):
         for tokenizer_class in [BertTokenizer, AutoTokenizer]:
             with self.assertRaises(EnvironmentError):
                 _ = tokenizer_class.from_pretrained("julien-c/herlolip-not-exists")
+
+    def test_parents_and_children_in_mappings(self):
+        # Test that the children are placed before the parents in the mappings, as the `instanceof` will be triggered
+        # by the parents and will return the wrong configuration type when using auto models
+
+        mappings = (TOKENIZER_MAPPING,)
+
+        for mapping in mappings:
+            mapping = tuple(mapping.items())
+            for index, (child_config, child_model) in enumerate(mapping[1:]):
+                for parent_config, parent_model in mapping[: index + 1]:
+                    with self.subTest(
+                        msg="Testing if {} is child of {}".format(child_config.__name__, parent_config.__name__)
+                    ):
+                        self.assertFalse(issubclass(child_config, parent_config))
+                        self.assertFalse(issubclass(child_model, parent_model))