update doc and tests

docs/source/index.rst

@@ -40,6 +40,7 @@ The library currently contains PyTorch implementations, pre-trained model weight
     :maxdepth: 2
     :caption: Package Reference
 
+    model_doc/auto
     model_doc/bert
     model_doc/gpt
     model_doc/transformerxl

docs/source/model_doc/auto.rst

@@ -1,9 +1,15 @@
-AutoModel, AutoConfig and AutoTokenizer - Standard derived classes
----------------------------------------------------------------
+AutoModels
+-----------
 
-In many case, the architecture you want to use can be guessed from the name or the path of the pretrained model you are supplying to the ``from_pretrained`` method.
+In many cases, the architecture you want to use can be guessed from the name or the path of the pretrained model you are supplying to the ``from_pretrained`` method.
+
+AutoClasses are here to do this job for you so that you automatically retreive the relevant model given the name/path to the pretrained weights/config/vocabulary.
+
+There are two types of AutoClasses:
+
+- ``AutoModel``, ``AutoConfig`` and ``AutoTokenizer``: instantiating these ones will directly create a class of the relevant architecture (ex: ``model = AutoModel.from_pretrained('bert-base-cased')`` will create a instance of ``BertModel``)
+- All the others (``AutoModelWithLMHead``, ``AutoModelForSequenceClassification``...)  are standardized Auto classes for finetuning. Instantiating these will create instance of the same class (``AutoModelWithLMHead``, ``AutoModelForSequenceClassification``...) comprising (i) the relevant base model class (as mentioned just above) and (ii) a standard fine-tuning head on top, convenient for the task.
 
-Auto classes are here to do this job for you so that you automatically retreive the relevant model given the name/path to the pretrained weights/config/vocabulary.
 
 ``AutoConfig``
 ~~~~~~~~~~~~~~~~~~~~~
@@ -19,6 +25,20 @@ Auto classes are here to do this job for you so that you automatically retreive
     :members:
 
 
+``AutoModelWithLMHead``
+~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.AutoModelWithLMHead
+    :members:
+
+
+``AutoModelForSequenceClassification``
+~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
+
+.. autoclass:: pytorch_transformers.AutoModelForSequenceClassification
+    :members:
+
+
 ``AutoTokenizer``
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~