AutoConfig + other Auto classes honor model_type

src/transformers/configuration_auto.py

@@ -16,6 +16,7 @@
 
 
 import logging
+from collections import OrderedDict
 
 from .configuration_albert import ALBERT_PRETRAINED_CONFIG_ARCHIVE_MAP, AlbertConfig
 from .configuration_bert import BERT_PRETRAINED_CONFIG_ARCHIVE_MAP, BertConfig
@@ -27,6 +28,7 @@ from .configuration_openai import OPENAI_GPT_PRETRAINED_CONFIG_ARCHIVE_MAP, Open
 from .configuration_roberta import ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, RobertaConfig
 from .configuration_t5 import T5_PRETRAINED_CONFIG_ARCHIVE_MAP, T5Config
 from .configuration_transfo_xl import TRANSFO_XL_PRETRAINED_CONFIG_ARCHIVE_MAP, TransfoXLConfig
+from .configuration_utils import PretrainedConfig
 from .configuration_xlm import XLM_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMConfig
 from .configuration_xlm_roberta import XLM_ROBERTA_PRETRAINED_CONFIG_ARCHIVE_MAP, XLMRobertaConfig
 from .configuration_xlnet import XLNET_PRETRAINED_CONFIG_ARCHIVE_MAP, XLNetConfig
@@ -56,17 +58,38 @@ ALL_PRETRAINED_CONFIG_ARCHIVE_MAP = dict(
 )
 
 
-class AutoConfig(object):
+CONFIG_MAPPING = OrderedDict(
+    [
+        ("t5", T5Config,),
+        ("distilbert", DistilBertConfig,),
+        ("albert", AlbertConfig,),
+        ("camembert", CamembertConfig,),
+        ("xlm-roberta", XLMRobertaConfig,),
+        ("roberta", RobertaConfig,),
+        ("bert", BertConfig,),
+        ("openai-gpt", OpenAIGPTConfig,),
+        ("gpt2", GPT2Config,),
+        ("transfo-xl", TransfoXLConfig,),
+        ("xlnet", XLNetConfig,),
+        ("xlm", XLMConfig,),
+        ("ctrl", CTRLConfig,),
+    ]
+)
+
+
+class AutoConfig:
     r""":class:`~transformers.AutoConfig` is a generic configuration class
         that will be instantiated as one of the configuration classes of the library
         when created with the `AutoConfig.from_pretrained(pretrained_model_name_or_path)`
         class method.
 
         The `from_pretrained()` method take care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
-        The base model class to instantiate is selected as the first pattern matching
+        When using string matching, the configuration class is matched on
-        in the `pretrained_model_name_or_path` string (in the following order):
+        the `pretrained_model_name_or_path` string in the following order:
+            - contains `t5`: T5Config (T5 model)
             - contains `distilbert`: DistilBertConfig (DistilBERT model)
             - contains `albert`: AlbertConfig (ALBERT model)
             - contains `camembert`: CamembertConfig (CamemBERT model)
@@ -90,41 +113,23 @@ class AutoConfig(object):
 
     @classmethod
     def for_model(cls, model_type, *args, **kwargs):
-        if "distilbert" in model_type:
+        for pattern, config_class in CONFIG_MAPPING.items():
-            return DistilBertConfig(*args, **kwargs)
+            if pattern in model_type:
-        elif "roberta" in model_type:
+                return config_class(*args, **kwargs)
-            return RobertaConfig(*args, **kwargs)
-        elif "bert" in model_type:
-            return BertConfig(*args, **kwargs)
-        elif "openai-gpt" in model_type:
-            return OpenAIGPTConfig(*args, **kwargs)
-        elif "gpt2" in model_type:
-            return GPT2Config(*args, **kwargs)
-        elif "transfo-xl" in model_type:
-            return TransfoXLConfig(*args, **kwargs)
-        elif "xlnet" in model_type:
-            return XLNetConfig(*args, **kwargs)
-        elif "xlm" in model_type:
-            return XLMConfig(*args, **kwargs)
-        elif "ctrl" in model_type:
-            return CTRLConfig(*args, **kwargs)
-        elif "albert" in model_type:
-            return AlbertConfig(*args, **kwargs)
-        elif "camembert" in model_type:
-            return CamembertConfig(*args, **kwargs)
         raise ValueError(
-            "Unrecognized model identifier in {}. Should contains one of "
+            "Unrecognized model identifier in {}. Should contain one of {}".format(
-            "'distilbert', 'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
+                model_type, ", ".join(CONFIG_MAPPING.keys())
-            "'xlm', 'roberta', 'ctrl', 'camembert', 'albert'".format(model_type)
+            )
         )
 
     @classmethod
     def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
-        r""" Instantiate a one of the configuration classes of the library
+        r""" Instantiate one of the configuration classes of the library
         from a pre-trained model configuration.
 
-        The configuration class to instantiate is selected as the first pattern matching
+        The configuration class to instantiate is selected
-        in the `pretrained_model_name_or_path` string (in the following order):
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
             - contains `t5`: T5Config (T5 model)
             - contains `distilbert`: DistilBertConfig (DistilBERT model)
             - contains `albert`: AlbertConfig (ALBERT model)
@@ -183,36 +188,21 @@ class AutoConfig(object):
             assert unused_kwargs == {'foo': False}
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config_dict, _ = PretrainedConfig.resolved_config_dict(
-            return T5Config.from_pretrained(pretrained_model_name_or_path, **kwargs)
+            pretrained_model_name_or_path, pretrained_config_archive_map=ALL_PRETRAINED_CONFIG_ARCHIVE_MAP, **kwargs
-        elif "distilbert" in pretrained_model_name_or_path:
+        )
-            return DistilBertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if "model_type" in config_dict:
-            return AlbertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+            config_class = CONFIG_MAPPING[config_dict["model_type"]]
-        elif "camembert" in pretrained_model_name_or_path:
+            return config_class.from_dict(config_dict, **kwargs)
-            return CamembertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+        else:
-        elif "xlm-roberta" in pretrained_model_name_or_path:
+            # Fallback: use pattern matching on the string.
-            return XLMRobertaConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+            for pattern, config_class in CONFIG_MAPPING.items():
-        elif "roberta" in pretrained_model_name_or_path:
+                if pattern in pretrained_model_name_or_path:
-            return RobertaConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+                    return config_class.from_dict(config_dict, **kwargs)
-        elif "bert" in pretrained_model_name_or_path:
+
-            return BertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "openai-gpt" in pretrained_model_name_or_path:
-            return OpenAIGPTConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "gpt2" in pretrained_model_name_or_path:
-            return GPT2Config.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "transfo-xl" in pretrained_model_name_or_path:
-            return TransfoXLConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
-            return XLNetConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "xlm" in pretrained_model_name_or_path:
-            return XLMConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-        elif "ctrl" in pretrained_model_name_or_path:
-            return CTRLConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
         raise ValueError(
-            "Unrecognized model identifier in {}. Should contains one of "
+            "Unrecognized model identifier in {}. Should have a `model_type` key in its config.json, or contain one of {}".format(
-            "'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
+                pretrained_model_name_or_path, ", ".join(CONFIG_MAPPING.keys())
-            "'xlm-roberta', 'xlm', 'roberta', 'distilbert', 'camembert', 'ctrl', 'albert'".format(
-                pretrained_model_name_or_path
             )
         )

src/transformers/configuration_utils.py

@@ -20,6 +20,7 @@ import copy
 import json
 import logging
 import os
+from typing import Dict, Optional, Tuple
 
 from .file_utils import CONFIG_NAME, cached_path, hf_bucket_url, is_remote_url
 
@@ -36,7 +37,7 @@ class PretrainedConfig(object):
             It only affects the model's configuration.
 
         Class attributes (overridden by derived classes):
-            - ``pretrained_config_archive_map``: a python ``dict`` of with `short-cut-names` (string) as keys and `url` (string) of associated pretrained model configurations as values.
+            - ``pretrained_config_archive_map``: a python ``dict`` with `shortcut names` (string) as keys and `url` (string) of associated pretrained model configurations as values.
 
         Parameters:
             ``finetuning_task``: string, default `None`. Name of the task used to fine-tune the model. This can be used when converting from an original (TensorFlow or PyTorch) checkpoint.
@@ -154,14 +155,32 @@ class PretrainedConfig(object):
             assert unused_kwargs == {'foo': False}
 
         """
+        config_dict, kwargs = cls.resolved_config_dict(pretrained_model_name_or_path, **kwargs)
+        return cls.from_dict(config_dict, **kwargs)
+
+    @classmethod
+    def resolved_config_dict(
+        cls, pretrained_model_name_or_path: str, pretrained_config_archive_map: Optional[Dict] = None, **kwargs
+    ) -> Tuple[Dict, Dict]:
+        """
+        From a `pretrained_model_name_or_path`, resolve to a dictionary of parameters, to be used
+        for instantiating a Config using `from_dict`.
+
+        Parameters:
+            pretrained_config_archive_map: (`optional`) Dict:
+                A map of `shortcut names` to `url`.
+                By default, will use the current class attribute.
+        """
         cache_dir = kwargs.pop("cache_dir", None)
         force_download = kwargs.pop("force_download", False)
         resume_download = kwargs.pop("resume_download", False)
         proxies = kwargs.pop("proxies", None)
-        return_unused_kwargs = kwargs.pop("return_unused_kwargs", False)
 
-        if pretrained_model_name_or_path in cls.pretrained_config_archive_map:
+        if pretrained_config_archive_map is None:
-            config_file = cls.pretrained_config_archive_map[pretrained_model_name_or_path]
+            pretrained_config_archive_map = cls.pretrained_config_archive_map
+
+        if pretrained_model_name_or_path in pretrained_config_archive_map:
+            config_file = pretrained_config_archive_map[pretrained_model_name_or_path]
         elif os.path.isdir(pretrained_model_name_or_path):
             config_file = os.path.join(pretrained_model_name_or_path, CONFIG_NAME)
         elif os.path.isfile(pretrained_model_name_or_path) or is_remote_url(pretrained_model_name_or_path):
@@ -178,23 +197,20 @@ class PretrainedConfig(object):
                 proxies=proxies,
                 resume_download=resume_download,
             )
-            # Load config
+            # Load config dict
-            config = cls.from_json_file(resolved_config_file)
+            config_dict = cls._dict_from_json_file(resolved_config_file)
 
         except EnvironmentError:
-            if pretrained_model_name_or_path in cls.pretrained_config_archive_map:
+            if pretrained_model_name_or_path in pretrained_config_archive_map:
                 msg = "Couldn't reach server at '{}' to download pretrained model configuration file.".format(
                     config_file
                 )
             else:
                 msg = (
-                    "Model name '{}' was not found in model name list ({}). "
+                    "Model name '{}' was not found in model name list. "
                     "We assumed '{}' was a path or url to a configuration file named {} or "
                     "a directory containing such a file but couldn't find any such file at this path or url.".format(
-                        pretrained_model_name_or_path,
+                        pretrained_model_name_or_path, config_file, CONFIG_NAME,
-                        ", ".join(cls.pretrained_config_archive_map.keys()),
-                        config_file,
-                        CONFIG_NAME,
                     )
                 )
             raise EnvironmentError(msg)
@@ -212,6 +228,15 @@ class PretrainedConfig(object):
         else:
             logger.info("loading configuration file {} from cache at {}".format(config_file, resolved_config_file))
 
+        return config_dict, kwargs
+
+    @classmethod
+    def from_dict(cls, config_dict: Dict, **kwargs):
+        """Constructs a `Config` from a Python dictionary of parameters."""
+        return_unused_kwargs = kwargs.pop("return_unused_kwargs", False)
+
+        config = cls(**config_dict)
+
         if hasattr(config, "pruned_heads"):
             config.pruned_heads = dict((int(key), value) for key, value in config.pruned_heads.items())
 
@@ -231,17 +256,16 @@ class PretrainedConfig(object):
             return config
 
     @classmethod
-    def from_dict(cls, json_object):
+    def from_json_file(cls, json_file: str):
-        """Constructs a `Config` from a Python dictionary of parameters."""
+        """Constructs a `Config` from a json file of parameters."""
-        return cls(**json_object)
+        config_dict = cls._dict_from_json_file(json_file)
+        return cls(**config_dict)
 
     @classmethod
-    def from_json_file(cls, json_file):
+    def _dict_from_json_file(cls, json_file: str):
-        """Constructs a `Config` from a json file of parameters."""
         with open(json_file, "r", encoding="utf-8") as reader:
             text = reader.read()
-        dict_obj = json.loads(text)
+        return json.loads(text)
-        return cls(**dict_obj)
 
     def __eq__(self, other):
         return self.__dict__ == other.__dict__

src/transformers/modeling_auto.py

@@ -19,6 +19,7 @@ import logging
 
 from .configuration_auto import (
     AlbertConfig,
+    AutoConfig,
     BertConfig,
     CamembertConfig,
     CTRLConfig,
@@ -26,11 +27,13 @@ from .configuration_auto import (
     GPT2Config,
     OpenAIGPTConfig,
     RobertaConfig,
+    T5Config,
     TransfoXLConfig,
     XLMConfig,
     XLMRobertaConfig,
     XLNetConfig,
 )
+from .configuration_utils import PretrainedConfig
 from .modeling_albert import (
     ALBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
     AlbertForMaskedLM,
@@ -129,7 +132,8 @@ class AutoModel(object):
         or the `AutoModel.from_config(config)` class methods.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The base model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -286,32 +290,36 @@ class AutoModel(object):
             model = AutoModel.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return T5Model.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "distilbert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return DistilBertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if isinstance(config, T5Config):
-            return AlbertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return T5Model.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "camembert" in pretrained_model_name_or_path:
+        elif isinstance(config, DistilBertConfig):
-            return CamembertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return DistilBertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "xlm-roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
-            return XLMRobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return AlbertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, CamembertConfig):
-            return RobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return CamembertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "bert" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMRobertaConfig):
-            return BertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return XLMRobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "openai-gpt" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
-            return OpenAIGPTModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return RobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "gpt2" in pretrained_model_name_or_path:
+        elif isinstance(config, BertConfig):
-            return GPT2Model.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return BertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "transfo-xl" in pretrained_model_name_or_path:
+        elif isinstance(config, OpenAIGPTConfig):
-            return TransfoXLModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return OpenAIGPTModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, GPT2Config):
-            return XLNetModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return GPT2Model.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "xlm" in pretrained_model_name_or_path:
+        elif isinstance(config, TransfoXLConfig):
-            return XLMModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TransfoXLModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "ctrl" in pretrained_model_name_or_path:
+        elif isinstance(config, XLNetConfig):
-            return CTRLModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return XLNetModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, XLMConfig):
+            return XLMModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, CTRLConfig):
+            return CTRLModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
             "'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
@@ -329,7 +337,8 @@ class AutoModelWithLMHead(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -407,7 +416,8 @@ class AutoModelWithLMHead(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -484,32 +494,56 @@ class AutoModelWithLMHead(object):
             model = AutoModelWithLMHead.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return T5WithLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "distilbert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return DistilBertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if isinstance(config, T5Config):
-            return AlbertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return T5WithLMHeadModel.from_pretrained(
-        elif "camembert" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return CamembertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "xlm-roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, DistilBertConfig):
-            return XLMRobertaForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return DistilBertForMaskedLM.from_pretrained(
-        elif "roberta" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return RobertaForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "bert" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
-            return BertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return AlbertForMaskedLM.from_pretrained(
-        elif "openai-gpt" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return OpenAIGPTLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "gpt2" in pretrained_model_name_or_path:
+        elif isinstance(config, CamembertConfig):
-            return GPT2LMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return CamembertForMaskedLM.from_pretrained(
-        elif "transfo-xl" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TransfoXLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMRobertaConfig):
-            return XLNetLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return XLMRobertaForMaskedLM.from_pretrained(
-        elif "xlm" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return XLMWithLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "ctrl" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
-            return CTRLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return RobertaForMaskedLM.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, BertConfig):
+            return BertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, OpenAIGPTConfig):
+            return OpenAIGPTLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, GPT2Config):
+            return GPT2LMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, TransfoXLConfig):
+            return TransfoXLLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return XLNetLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLMConfig):
+            return XLMWithLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, CTRLConfig):
+            return CTRLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
             "'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
@@ -527,7 +561,8 @@ class AutoModelForSequenceClassification(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -592,7 +627,8 @@ class AutoModelForSequenceClassification(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -665,32 +701,42 @@ class AutoModelForSequenceClassification(object):
             model = AutoModelForSequenceClassification.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "distilbert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
+        if not isinstance(config, PretrainedConfig):
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+
+        if isinstance(config, DistilBertConfig):
             return DistilBertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "albert" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
             return AlbertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "camembert" in pretrained_model_name_or_path:
+        elif isinstance(config, CamembertConfig):
             return CamembertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "xlm-roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMRobertaConfig):
             return XLMRobertaForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
             return RobertaForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, BertConfig):
+            return BertForSequenceClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return XLNetForSequenceClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLMConfig):
+            return XLMForSequenceClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "bert" in pretrained_model_name_or_path:
-            return BertForSequenceClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
-            return XLNetForSequenceClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-        elif "xlm" in pretrained_model_name_or_path:
-            return XLMForSequenceClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
@@ -708,7 +754,8 @@ class AutoModelForQuestionAnswering(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -763,7 +810,8 @@ class AutoModelForQuestionAnswering(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -830,16 +878,30 @@ class AutoModelForQuestionAnswering(object):
             model = AutoModelForQuestionAnswering.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "distilbert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return DistilBertForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "albert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return AlbertForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "bert" in pretrained_model_name_or_path:
+        if isinstance(config, DistilBertConfig):
-            return BertForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return DistilBertForQuestionAnswering.from_pretrained(
-        elif "xlnet" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return XLNetForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "xlm" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
-            return XLMForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return AlbertForQuestionAnswering.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, BertConfig):
+            return BertForQuestionAnswering.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return XLNetForQuestionAnswering.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLMConfig):
+            return XLMForQuestionAnswering.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
@@ -893,7 +955,8 @@ class AutoModelForTokenClassification:
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -959,24 +1022,34 @@ class AutoModelForTokenClassification:
             model = AutoModelForTokenClassification.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "camembert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
+        if not isinstance(config, PretrainedConfig):
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+
+        if isinstance(config, CamembertConfig):
             return CamembertForTokenClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "distilbert" in pretrained_model_name_or_path:
+        elif isinstance(config, DistilBertConfig):
             return DistilBertForTokenClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "xlm-roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMRobertaConfig):
             return XLMRobertaForTokenClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, RobertaConfig):
+            return RobertaForTokenClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, BertConfig):
+            return BertForTokenClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return XLNetForTokenClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "roberta" in pretrained_model_name_or_path:
-            return RobertaForTokenClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-        elif "bert" in pretrained_model_name_or_path:
-            return BertForTokenClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
-            return XLNetForTokenClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "

src/transformers/modeling_tf_auto.py

@@ -18,16 +18,20 @@
 import logging
 
 from .configuration_auto import (
+    AlbertConfig,
+    AutoConfig,
     BertConfig,
     CTRLConfig,
     DistilBertConfig,
     GPT2Config,
     OpenAIGPTConfig,
     RobertaConfig,
+    T5Config,
     TransfoXLConfig,
     XLMConfig,
     XLNetConfig,
 )
+from .configuration_utils import PretrainedConfig
 from .modeling_tf_albert import (
     TF_ALBERT_PRETRAINED_MODEL_ARCHIVE_MAP,
     TFAlbertForMaskedLM,
@@ -113,7 +117,8 @@ class TFAutoModel(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The base model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -257,28 +262,38 @@ class TFAutoModel(object):
             model = TFAutoModel.from_pretrained('./pt_model/bert_pytorch_model.bin', from_pt=True, config=config)
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return TFT5Model.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "distilbert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return TFDistilBertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if isinstance(config, T5Config):
-            return TFAlbertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFT5Model.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, DistilBertConfig):
-            return TFRobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFDistilBertModel.from_pretrained(
-        elif "bert" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFBertModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "openai-gpt" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
-            return TFOpenAIGPTModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFAlbertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "gpt2" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
-            return TFGPT2Model.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFRobertaModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "transfo-xl" in pretrained_model_name_or_path:
+        elif isinstance(config, BertConfig):
-            return TFTransfoXLModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFBertModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, OpenAIGPTConfig):
-            return TFXLNetModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFOpenAIGPTModel.from_pretrained(
-        elif "xlm" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFXLMModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "ctrl" in pretrained_model_name_or_path:
+        elif isinstance(config, GPT2Config):
-            return TFCTRLModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFGPT2Model.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, TransfoXLConfig):
+            return TFTransfoXLModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return TFXLNetModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, XLMConfig):
+            return TFXLMModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
+        elif isinstance(config, CTRLConfig):
+            return TFCTRLModel.from_pretrained(pretrained_model_name_or_path, *model_args, config=config, **kwargs)
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
@@ -295,7 +310,8 @@ class TFAutoModelWithLMHead(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -368,7 +384,8 @@ class TFAutoModelWithLMHead(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -443,28 +460,54 @@ class TFAutoModelWithLMHead(object):
             model = TFAutoModelWithLMHead.from_pretrained('./pt_model/bert_pytorch_model.bin', from_pt=True, config=config)
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return TFT5WithLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "distilbert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return TFDistilBertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if isinstance(config, T5Config):
-            return TFAlbertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFT5WithLMHeadModel.from_pretrained(
-        elif "roberta" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFRobertaForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "bert" in pretrained_model_name_or_path:
+        elif isinstance(config, DistilBertConfig):
-            return TFBertForMaskedLM.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFDistilBertForMaskedLM.from_pretrained(
-        elif "openai-gpt" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFOpenAIGPTLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "gpt2" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
-            return TFGPT2LMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFAlbertForMaskedLM.from_pretrained(
-        elif "transfo-xl" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFTransfoXLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
-            return TFXLNetLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFRobertaForMaskedLM.from_pretrained(
-        elif "xlm" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
-            return TFXLMWithLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            )
-        elif "ctrl" in pretrained_model_name_or_path:
+        elif isinstance(config, BertConfig):
-            return TFCTRLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFBertForMaskedLM.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, OpenAIGPTConfig):
+            return TFOpenAIGPTLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, GPT2Config):
+            return TFGPT2LMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, TransfoXLConfig):
+            return TFTransfoXLLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
+            return TFXLNetLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLMConfig):
+            return TFXLMWithLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, CTRLConfig):
+            return TFCTRLLMHeadModel.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
@@ -481,7 +524,8 @@ class TFAutoModelForSequenceClassification(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -537,7 +581,8 @@ class TFAutoModelForSequenceClassification(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -610,28 +655,34 @@ class TFAutoModelForSequenceClassification(object):
             model = TFAutoModelForSequenceClassification.from_pretrained('./pt_model/bert_pytorch_model.bin', from_pt=True, config=config)
 
         """
-        if "distilbert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
+        if not isinstance(config, PretrainedConfig):
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+
+        if isinstance(config, DistilBertConfig):
             return TFDistilBertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "albert" in pretrained_model_name_or_path:
+        elif isinstance(config, AlbertConfig):
             return TFAlbertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, RobertaConfig):
             return TFRobertaForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "bert" in pretrained_model_name_or_path:
+        elif isinstance(config, BertConfig):
             return TFBertForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, XLNetConfig):
             return TFXLNetForSequenceClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLMConfig):
+            return TFXLMForSequenceClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "xlm" in pretrained_model_name_or_path:
-            return TFXLMForSequenceClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
 
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "
@@ -647,7 +698,8 @@ class TFAutoModelForQuestionAnswering(object):
         class method.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -699,7 +751,8 @@ class TFAutoModelForQuestionAnswering(object):
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -771,19 +824,25 @@ class TFAutoModelForQuestionAnswering(object):
             model = TFAutoModelForQuestionAnswering.from_pretrained('./pt_model/bert_pytorch_model.bin', from_pt=True, config=config)
 
         """
-        if "distilbert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
+        if not isinstance(config, PretrainedConfig):
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
+
+        if isinstance(config, DistilBertConfig):
             return TFDistilBertForQuestionAnswering.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "bert" in pretrained_model_name_or_path:
+        elif isinstance(config, BertConfig):
-            return TFBertForQuestionAnswering.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+            return TFBertForQuestionAnswering.from_pretrained(
-        elif "xlnet" in pretrained_model_name_or_path:
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, XLNetConfig):
             return TFXLNetForQuestionAnsweringSimple.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "xlm" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMConfig):
             return TFXLMForQuestionAnsweringSimple.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
 
         raise ValueError(
@@ -833,7 +892,8 @@ class TFAutoModelForTokenClassification:
         from a pre-trained model configuration.
 
         The `from_pretrained()` method takes care of returning the correct model class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The model class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -898,17 +958,25 @@ class TFAutoModelForTokenClassification:
             model = TFAutoModelForTokenClassification.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
 
         """
-        if "bert" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return TFBertForTokenClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "xlnet" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return TFXLNetForTokenClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
+
-        elif "distilbert" in pretrained_model_name_or_path:
+        if isinstance(config, BertConfig):
-            return TFDistilBertForTokenClassification.from_pretrained(
+            return TFBertForTokenClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
-        elif "roberta" in pretrained_model_name_or_path:
+        elif isinstance(config, XLNetConfig):
+            return TFXLNetForTokenClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, DistilBertConfig):
+            return TFDistilBertForTokenClassification.from_pretrained(
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
+            )
+        elif isinstance(config, RobertaConfig):
             return TFRobertaForTokenClassification.from_pretrained(
-                pretrained_model_name_or_path, *model_args, **kwargs
+                pretrained_model_name_or_path, *model_args, config=config, **kwargs
             )
 
         raise ValueError(

src/transformers/tokenization_auto.py

@@ -17,6 +17,23 @@
 
 import logging
 
+from .configuration_auto import (
+    AlbertConfig,
+    AutoConfig,
+    BertConfig,
+    CamembertConfig,
+    CTRLConfig,
+    DistilBertConfig,
+    GPT2Config,
+    OpenAIGPTConfig,
+    RobertaConfig,
+    T5Config,
+    TransfoXLConfig,
+    XLMConfig,
+    XLMRobertaConfig,
+    XLNetConfig,
+)
+from .configuration_utils import PretrainedConfig
 from .tokenization_albert import AlbertTokenizer
 from .tokenization_bert import BertTokenizer
 from .tokenization_bert_japanese import BertJapaneseTokenizer
@@ -43,7 +60,8 @@ class AutoTokenizer(object):
         class method.
 
         The `from_pretrained()` method take care of returning the correct tokenizer class instance
-        using pattern matching on the `pretrained_model_name_or_path` string.
+        based on the `model_type` property of the config object, or when it's missing,
+        falling back to using pattern matching on the `pretrained_model_name_or_path` string.
 
         The tokenizer class to instantiate is selected as the first pattern matching
         in the `pretrained_model_name_or_path` string (in the following order):
@@ -72,7 +90,7 @@ class AutoTokenizer(object):
 
     @classmethod
     def from_pretrained(cls, pretrained_model_name_or_path, *inputs, **kwargs):
-        r""" Instantiate a one of the tokenizer classes of the library
+        r""" Instantiate one of the tokenizer classes of the library
         from a pre-trained model vocabulary.
 
         The tokenizer class to instantiate is selected as the first pattern matching
@@ -129,33 +147,38 @@ class AutoTokenizer(object):
             tokenizer = AutoTokenizer.from_pretrained('./test/bert_saved_model/')
 
         """
-        if "t5" in pretrained_model_name_or_path:
+        config = kwargs.pop("config", None)
-            return T5Tokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        if not isinstance(config, PretrainedConfig):
-        elif "distilbert" in pretrained_model_name_or_path:
+            config = AutoConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
-            return DistilBertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+
-        elif "albert" in pretrained_model_name_or_path:
+        if "bert-base-japanese" in pretrained_model_name_or_path:
-            return AlbertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "camembert" in pretrained_model_name_or_path:
-            return CamembertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "xlm-roberta" in pretrained_model_name_or_path:
-            return XLMRobertaTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "roberta" in pretrained_model_name_or_path:
-            return RobertaTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "bert-base-japanese" in pretrained_model_name_or_path:
             return BertJapaneseTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "bert" in pretrained_model_name_or_path:
+
+        if isinstance(config, T5Config):
+            return T5Tokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, DistilBertConfig):
+            return DistilBertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, AlbertConfig):
+            return AlbertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, CamembertConfig):
+            return CamembertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, XLMRobertaConfig):
+            return XLMRobertaTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, RobertaConfig):
+            return RobertaTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
+        elif isinstance(config, BertConfig):
             return BertTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "openai-gpt" in pretrained_model_name_or_path:
+        elif isinstance(config, OpenAIGPTConfig):
             return OpenAIGPTTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "gpt2" in pretrained_model_name_or_path:
+        elif isinstance(config, GPT2Config):
             return GPT2Tokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "transfo-xl" in pretrained_model_name_or_path:
+        elif isinstance(config, TransfoXLConfig):
             return TransfoXLTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "xlnet" in pretrained_model_name_or_path:
+        elif isinstance(config, XLNetConfig):
             return XLNetTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "xlm" in pretrained_model_name_or_path:
+        elif isinstance(config, XLMConfig):
             return XLMTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
-        elif "ctrl" in pretrained_model_name_or_path:
+        elif isinstance(config, CTRLConfig):
             return CTRLTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
         raise ValueError(
             "Unrecognized model identifier in {}. Should contains one of "

tests/fixtures/dummy-config.json

@@ -0,0 +1,3 @@
+{
+  "model_type": "roberta"
+}

tests/test_configuration_auto.py

@@ -0,0 +1,38 @@
+# coding=utf-8
+# Copyright 2019-present, the HuggingFace Inc. team.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# 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.
+
+import os
+import unittest
+
+from transformers.configuration_auto import AutoConfig
+from transformers.configuration_bert import BertConfig
+from transformers.configuration_roberta import RobertaConfig
+
+
+SAMPLE_ROBERTA_CONFIG = os.path.join(os.path.dirname(os.path.abspath(__file__)), "fixtures/dummy-config.json")
+
+
+class AutoConfigTest(unittest.TestCase):
+    def test_config_from_model_shortcut(self):
+        config = AutoConfig.from_pretrained("bert-base-uncased")
+        self.assertIsInstance(config, BertConfig)
+
+    def test_config_from_model_type(self):
+        config = AutoConfig.from_pretrained(SAMPLE_ROBERTA_CONFIG)
+        self.assertIsInstance(config, RobertaConfig)
+
+    def test_config_for_model_str(self):
+        config = AutoConfig.for_model("roberta")
+        self.assertIsInstance(config, RobertaConfig)

tests/test_modeling_auto.py

@@ -83,7 +83,7 @@ class AutoModelTest(unittest.TestCase):
             self.assertIsNotNone(model)
             self.assertIsInstance(model, BertForSequenceClassification)
 
-    @slow
+    # @slow
     def test_question_answering_model_from_pretrained(self):
         logging.basicConfig(level=logging.INFO)
         for model_name in list(BERT_PRETRAINED_MODEL_ARCHIVE_MAP.keys())[:1]:

tests/test_tokenization_auto.py

@@ -29,7 +29,7 @@ from .utils import SMALL_MODEL_IDENTIFIER, slow
 
 
 class AutoTokenizerTest(unittest.TestCase):
-    @slow
+    # @slow
     def test_tokenizer_from_pretrained(self):
         logging.basicConfig(level=logging.INFO)
         for model_name in list(BERT_PRETRAINED_CONFIG_ARCHIVE_MAP.keys())[:1]: