updating AutoModels and AutoConfiguration - adding pipelines
This commit is contained in:
thomwolf
committed by
Morgan Funtowicz
Morgan Funtowicz
parent
2d8559731a
commit
b81ab431f2
@@ -61,6 +61,34 @@ class AutoConfig(object):
|
||||
raise EnvironmentError("AutoConfig is designed to be instantiated "
|
||||
"using the `AutoConfig.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
|
||||
@classmethod
|
||||
def for_model(cls, model_type, *args, **kwargs):
|
||||
if 'distilbert' in model_type:
|
||||
return DistilBertConfig(*args, **kwargs)
|
||||
elif 'roberta' in model_type:
|
||||
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 "
|
||||
"'distilbert', 'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
|
||||
"'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
|
||||
|
||||
@@ -18,6 +18,10 @@ from __future__ import absolute_import, division, print_function, unicode_litera
|
||||
|
||||
import logging
|
||||
|
||||
from .configuration_auto import (AlbertConfig, BertConfig, CamembertConfig, CTRLConfig,
|
||||
DistilBertConfig, GPT2Config, OpenAIGPTConfig, RobertaConfig,
|
||||
TransfoXLConfig, XLMConfig, XLNetConfig)
|
||||
|
||||
from .modeling_bert import BertModel, BertForMaskedLM, BertForSequenceClassification, BertForQuestionAnswering
|
||||
from .modeling_openai import OpenAIGPTModel, OpenAIGPTLMHeadModel
|
||||
from .modeling_gpt2 import GPT2Model, GPT2LMHeadModel
|
||||
@@ -27,8 +31,7 @@ from .modeling_xlnet import XLNetModel, XLNetLMHeadModel, XLNetForSequenceClassi
|
||||
from .modeling_xlm import XLMModel, XLMWithLMHeadModel, XLMForSequenceClassification, XLMForQuestionAnswering
|
||||
from .modeling_roberta import RobertaModel, RobertaForMaskedLM, RobertaForSequenceClassification
|
||||
from .modeling_distilbert import DistilBertModel, DistilBertForQuestionAnswering, DistilBertForMaskedLM, DistilBertForSequenceClassification
|
||||
from .modeling_camembert import CamembertModel, CamembertForMaskedLM, CamembertForSequenceClassification, CamembertForMultipleChoice
|
||||
from .modeling_camembert import CamembertModel, CamembertForMaskedLM, CamembertForSequenceClassification, CamembertForMultipleChoice
|
||||
from .modeling_camembert import CamembertModel, CamembertForQuestionAnswering, CamembertForMaskedLM, CamembertForSequenceClassification, CamembertForMultipleChoice
|
||||
from .modeling_albert import AlbertModel, AlbertForMaskedLM, AlbertForSequenceClassification, AlbertForQuestionAnswering
|
||||
|
||||
from .modeling_utils import PreTrainedModel, SequenceSummary
|
||||
@@ -43,7 +46,7 @@ class AutoModel(object):
|
||||
:class:`~transformers.AutoModel` is a generic model class
|
||||
that will be instantiated as one of the base model classes of the library
|
||||
when created with the `AutoModel.from_pretrained(pretrained_model_name_or_path)`
|
||||
class method.
|
||||
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.
|
||||
@@ -66,7 +69,54 @@ class AutoModel(object):
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("AutoModel is designed to be instantiated "
|
||||
"using the `AutoModel.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
"using the `AutoModel.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`AutoModel.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: RobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `openai-gpt` configuration class: OpenAIGPTModel (OpenAI GPT model)
|
||||
- isInstance of `gpt2` configuration class: GPT2Model (OpenAI GPT-2 model)
|
||||
- isInstance of `ctrl` configuration class: CTRLModel (Salesforce CTRL model)
|
||||
- isInstance of `transfo-xl` configuration class: TransfoXLModel (Transformer-XL model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModel.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return DistilBertModel(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return RobertaModel(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return BertModel(config)
|
||||
elif isinstance(config, OpenAIGPTConfig):
|
||||
return OpenAIGPTModel(config)
|
||||
elif isinstance(config, GPT2Config):
|
||||
return GPT2Model(config)
|
||||
elif isinstance(config, TransfoXLConfig):
|
||||
return TransfoXLModel(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return XLNetModel(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return XLMModel(config)
|
||||
elif isinstance(config, CTRLConfig):
|
||||
return CTRLModel(config)
|
||||
elif isinstance(config, AlbertConfig):
|
||||
return AlbertModel(config)
|
||||
elif isinstance(config, CamembertConfig):
|
||||
return CamembertModel(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -201,7 +251,54 @@ class AutoModelWithLMHead(object):
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("AutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `AutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
"using the `AutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`AutoModelWithLMHead.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: RobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `openai-gpt` configuration class: OpenAIGPTModel (OpenAI GPT model)
|
||||
- isInstance of `gpt2` configuration class: GPT2Model (OpenAI GPT-2 model)
|
||||
- isInstance of `ctrl` configuration class: CTRLModel (Salesforce CTRL model)
|
||||
- isInstance of `transfo-xl` configuration class: TransfoXLModel (Transformer-XL model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelWithLMHead.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return DistilBertForMaskedLM(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return RobertaForMaskedLM(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return BertForMaskedLM(config)
|
||||
elif isinstance(config, OpenAIGPTConfig):
|
||||
return OpenAIGPTLMHeadModel(config)
|
||||
elif isinstance(config, GPT2Config):
|
||||
return GPT2LMHeadModel(config)
|
||||
elif isinstance(config, TransfoXLConfig):
|
||||
return TransfoXLLMHeadModel(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return XLNetLMHeadModel(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return XLMWithLMHeadModel(config)
|
||||
elif isinstance(config, CTRLConfig):
|
||||
return CTRLLMHeadModel(config)
|
||||
elif isinstance(config, AlbertConfig):
|
||||
return AlbertLMHeadModel(config)
|
||||
elif isinstance(config, CamembertConfig):
|
||||
return CamembertLMHeadModel(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -333,8 +430,43 @@ class AutoModelForSequenceClassification(object):
|
||||
This class cannot be instantiated using `__init__()` (throws an error).
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("AutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `AutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
raise EnvironmentError("AutoModelForSequenceClassification is designed to be instantiated "
|
||||
"using the `AutoModelForSequenceClassification.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`AutoModelForSequenceClassification.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: RobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelForSequenceClassification.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, AlbertConfig):
|
||||
return AlbertForSequenceClassification(config)
|
||||
elif isintance(config, CamembertConfig):
|
||||
return CamembertForSequenceClassification(config)
|
||||
elif isinstance(config, DistilBertConfig):
|
||||
return DistilBertForSequenceClassification(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return RobertaForSequenceClassification(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return BertForSequenceClassification(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return XLNetForSequenceClassification(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return XLMForSequenceClassification(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -453,8 +585,40 @@ class AutoModelForQuestionAnswering(object):
|
||||
This class cannot be instantiated using `__init__()` (throws an error).
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("AutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `AutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
raise EnvironmentError("AutoModelForQuestionAnswering is designed to be instantiated "
|
||||
"using the `AutoModelForQuestionAnswering.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`AutoModelForQuestionAnswering.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelForSequenceClassification.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isintance(config, AlbertConfig):
|
||||
return AlbertForQuestionAnswering(config)
|
||||
elif isintance(config, CamembertConfig):
|
||||
return CamembertForQuestionAnswering(config)
|
||||
elif isinstance(config, DistilBertConfig):
|
||||
return DistilBertForQuestionAnswering(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return BertForQuestionAnswering(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return XLNetForQuestionAnswering(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return XLMForQuestionAnswering(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
|
||||
@@ -18,6 +18,10 @@ from __future__ import absolute_import, division, print_function, unicode_litera
|
||||
|
||||
import logging
|
||||
|
||||
from .configuration_auto import (BertConfig, CTRLConfig, DistilBertConfig,
|
||||
GPT2Config, OpenAIGPTConfig, RobertaConfig,
|
||||
TransfoXLConfig, XLMConfig, XLNetConfig)
|
||||
|
||||
from .modeling_tf_bert import TFBertModel, TFBertForMaskedLM, TFBertForSequenceClassification, TFBertForQuestionAnswering
|
||||
from .modeling_tf_openai import TFOpenAIGPTModel, TFOpenAIGPTLMHeadModel
|
||||
from .modeling_tf_gpt2 import TFGPT2Model, TFGPT2LMHeadModel
|
||||
@@ -59,7 +63,50 @@ class TFAutoModel(object):
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("TFAutoModel is designed to be instantiated "
|
||||
"using the `TFAutoModel.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
"using the `TFAutoModel.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`TFAutoModel.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: TFDistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: TFRobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: TFBertModel (Bert model)
|
||||
- isInstance of `openai-gpt` configuration class: TFOpenAIGPTModel (OpenAI GPT model)
|
||||
- isInstance of `gpt2` configuration class: TFGPT2Model (OpenAI GPT-2 model)
|
||||
- isInstance of `ctrl` configuration class: TFCTRLModel (Salesforce CTRL model)
|
||||
- isInstance of `transfo-xl` configuration class: TFTransfoXLModel (Transformer-XL model)
|
||||
- isInstance of `xlnet` configuration class: TFXLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: TFXLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = TFAutoModel.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return TFDistilBertModel(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return TFRobertaModel(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return TFBertModel(config)
|
||||
elif isinstance(config, OpenAIGPTConfig):
|
||||
return TFOpenAIGPTModel(config)
|
||||
elif isinstance(config, GPT2Config):
|
||||
return TFGPT2Model(config)
|
||||
elif isinstance(config, TransfoXLConfig):
|
||||
return TFTransfoXLModel(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return TFXLNetModel(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return TFXLMModel(config)
|
||||
elif isinstance(config, CTRLConfig):
|
||||
return TFCTRLModel(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -156,7 +203,7 @@ class TFAutoModel(object):
|
||||
return TFCTRLModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
|
||||
|
||||
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
|
||||
"'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
|
||||
"'distilbert', 'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
|
||||
"'xlm', 'roberta', 'ctrl'".format(pretrained_model_name_or_path))
|
||||
|
||||
|
||||
@@ -186,7 +233,50 @@ class TFAutoModelWithLMHead(object):
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("TFAutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `TFAutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
"using the `TFAutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`TFAutoModelWithLMHead.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: RobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `openai-gpt` configuration class: OpenAIGPTModel (OpenAI GPT model)
|
||||
- isInstance of `gpt2` configuration class: GPT2Model (OpenAI GPT-2 model)
|
||||
- isInstance of `ctrl` configuration class: CTRLModel (Salesforce CTRL model)
|
||||
- isInstance of `transfo-xl` configuration class: TransfoXLModel (Transformer-XL model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelWithLMHead.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return TFDistilBertForMaskedLM(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return TFRobertaForMaskedLM(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return TFBertForMaskedLM(config)
|
||||
elif isinstance(config, OpenAIGPTConfig):
|
||||
return TFOpenAIGPTLMHeadModel(config)
|
||||
elif isinstance(config, GPT2Config):
|
||||
return TFGPT2LMHeadModel(config)
|
||||
elif isinstance(config, TransfoXLConfig):
|
||||
return TFTransfoXLLMHeadModel(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return TFXLNetLMHeadModel(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return TFXLMWithLMHeadModel(config)
|
||||
elif isinstance(config, CTRLConfig):
|
||||
return TFCTRLLMHeadModel(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -287,7 +377,7 @@ class TFAutoModelWithLMHead(object):
|
||||
return TFCTRLLMHeadModel.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
|
||||
|
||||
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
|
||||
"'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
|
||||
"'distilbert', 'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
|
||||
"'xlm', 'roberta', 'ctrl'".format(pretrained_model_name_or_path))
|
||||
|
||||
|
||||
@@ -312,8 +402,39 @@ class TFAutoModelForSequenceClassification(object):
|
||||
This class cannot be instantiated using `__init__()` (throws an error).
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("TFAutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `TFAutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
raise EnvironmentError("TFAutoModelForSequenceClassification is designed to be instantiated "
|
||||
"using the `TFAutoModelForSequenceClassification.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`TFAutoModelForSequenceClassification.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `roberta` configuration class: RobertaModel (RoBERTa model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelForSequenceClassification.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return TFDistilBertForSequenceClassification(config)
|
||||
elif isinstance(config, RobertaConfig):
|
||||
return TFRobertaForSequenceClassification(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return TFBertForSequenceClassification(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return TFXLNetForSequenceClassification(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return TFXLMForSequenceClassification(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -405,7 +526,7 @@ class TFAutoModelForSequenceClassification(object):
|
||||
return TFXLMForSequenceClassification.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
|
||||
|
||||
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
|
||||
"'bert', 'xlnet', 'xlm', 'roberta'".format(pretrained_model_name_or_path))
|
||||
"'distilbert', 'bert', 'xlnet', 'xlm', 'roberta'".format(pretrained_model_name_or_path))
|
||||
|
||||
|
||||
class TFAutoModelForQuestionAnswering(object):
|
||||
@@ -428,8 +549,36 @@ class TFAutoModelForQuestionAnswering(object):
|
||||
This class cannot be instantiated using `__init__()` (throws an error).
|
||||
"""
|
||||
def __init__(self):
|
||||
raise EnvironmentError("TFAutoModelWithLMHead is designed to be instantiated "
|
||||
"using the `TFAutoModelWithLMHead.from_pretrained(pretrained_model_name_or_path)` method.")
|
||||
raise EnvironmentError("TFAutoModelForQuestionAnswering is designed to be instantiated "
|
||||
"using the `TFAutoModelForQuestionAnswering.from_pretrained(pretrained_model_name_or_path)` or "
|
||||
"`TFAutoModelForQuestionAnswering.from_config(config)` methods.")
|
||||
|
||||
@classmethod
|
||||
def from_config(cls, config):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a configuration.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
The model class to instantiate is selected based on the configuration class:
|
||||
- isInstance of `distilbert` configuration class: DistilBertModel (DistilBERT model)
|
||||
- isInstance of `bert` configuration class: BertModel (Bert model)
|
||||
- isInstance of `xlnet` configuration class: XLNetModel (XLNet model)
|
||||
- isInstance of `xlm` configuration class: XLMModel (XLM model)
|
||||
|
||||
Examples::
|
||||
|
||||
config = BertConfig.from_pretrained('bert-base-uncased') # Download configuration from S3 and cache.
|
||||
model = AutoModelForSequenceClassification.from_config(config) # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
"""
|
||||
if isinstance(config, DistilBertConfig):
|
||||
return TFDistilBertForQuestionAnswering(config)
|
||||
elif isinstance(config, BertConfig):
|
||||
return TFBertForQuestionAnswering(config)
|
||||
elif isinstance(config, XLNetConfig):
|
||||
return TFXLNetForQuestionAnswering(config)
|
||||
elif isinstance(config, XLMConfig):
|
||||
return TFXLMForQuestionAnswering(config)
|
||||
raise ValueError("Unrecognized configuration class {}".format(config))
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, *model_args, **kwargs):
|
||||
@@ -518,4 +667,4 @@ class TFAutoModelForQuestionAnswering(object):
|
||||
return TFXLMForQuestionAnsweringSimple.from_pretrained(pretrained_model_name_or_path, *model_args, **kwargs)
|
||||
|
||||
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
|
||||
"'bert', 'xlnet', 'xlm'".format(pretrained_model_name_or_path))
|
||||
"'distilbert', 'bert', 'xlnet', 'xlm'".format(pretrained_model_name_or_path))
|
||||
|
||||
+12
-71
@@ -58,7 +58,7 @@ class TextClassificationPipeline(object):
|
||||
- contains `xlnet`: XLNetModel (XLNet model)
|
||||
- contains `xlm`: XLMModel (XLM model)
|
||||
"""
|
||||
def __init__(self, tokenizer, model):
|
||||
def __init__(self, tokenizer, model, is_compiled=False, is_trained=False):
|
||||
self.tokenizer = tokenizer
|
||||
self.model = model
|
||||
if is_tf_available():
|
||||
@@ -67,78 +67,13 @@ class TextClassificationPipeline(object):
|
||||
self.framework = 'pt'
|
||||
else:
|
||||
raise ImportError("At least one of PyTorch or TensorFlow 2.0+ should be installed to use CLI training")
|
||||
self.is_compiled = False
|
||||
self.is_compiled = is_compiled
|
||||
self.is_trained = is_trained
|
||||
|
||||
|
||||
@classmethod
|
||||
def from_pretrained(cls, pretrained_model_name_or_path, **kwargs):
|
||||
r""" Instantiates one of the base model classes of the library
|
||||
from a pre-trained model configuration.
|
||||
|
||||
The model class to instantiate is selected as the first pattern matching
|
||||
in the `pretrained_model_name_or_path` string (in the following order):
|
||||
- contains `distilbert`: DistilBertModel (DistilBERT model)
|
||||
- contains `roberta`: RobertaModel (RoBERTa model)
|
||||
- contains `bert`: BertModel (Bert model)
|
||||
- contains `openai-gpt`: OpenAIGPTModel (OpenAI GPT model)
|
||||
- contains `gpt2`: GPT2Model (OpenAI GPT-2 model)
|
||||
- contains `ctrl`: CTRLModel (Salesforce CTRL model)
|
||||
- contains `transfo-xl`: TransfoXLModel (Transformer-XL model)
|
||||
- contains `xlnet`: XLNetModel (XLNet model)
|
||||
- contains `xlm`: XLMModel (XLM model)
|
||||
|
||||
The model is set in evaluation mode by default using `model.eval()` (Dropout modules are deactivated)
|
||||
To train the model, you should first set it back in training mode with `model.train()`
|
||||
|
||||
Params:
|
||||
pretrained_model_name_or_path: either:
|
||||
|
||||
- a string with the `shortcut name` of a pre-trained model to load from cache or download, e.g.: ``bert-base-uncased``.
|
||||
- a path to a `directory` containing model weights saved using :func:`~transformers.PreTrainedModel.save_pretrained`, e.g.: ``./my_model_directory/``.
|
||||
- a path or url to a `tensorflow index checkpoint file` (e.g. `./tf_model/model.ckpt.index`). In this case, ``from_tf`` should be set to True and a configuration object should be provided as ``config`` argument. This loading path is slower than converting the TensorFlow checkpoint in a PyTorch model using the provided conversion scripts and loading the PyTorch model afterwards.
|
||||
|
||||
config: (`optional`) instance of a class derived from :class:`~transformers.PretrainedConfig`:
|
||||
Configuration for the model to use instead of an automatically loaded configuation. Configuration can be automatically loaded when:
|
||||
|
||||
- the model is a model provided by the library (loaded with the ``shortcut-name`` string of a pretrained model), or
|
||||
- the model was saved using :func:`~transformers.PreTrainedModel.save_pretrained` and is reloaded by suppling the save directory.
|
||||
- the model is loaded by suppling a local directory as ``pretrained_model_name_or_path`` and a configuration JSON file named `config.json` is found in the directory.
|
||||
|
||||
state_dict: (`optional`) dict:
|
||||
an optional state dictionnary for the model to use instead of a state dictionary loaded from saved weights file.
|
||||
This option can be used if you want to create a model from a pretrained configuration but load your own weights.
|
||||
In this case though, you should check if using :func:`~transformers.PreTrainedModel.save_pretrained` and :func:`~transformers.PreTrainedModel.from_pretrained` is not a simpler option.
|
||||
|
||||
cache_dir: (`optional`) string:
|
||||
Path to a directory in which a downloaded pre-trained model
|
||||
configuration should be cached if the standard cache should not be used.
|
||||
|
||||
force_download: (`optional`) boolean, default False:
|
||||
Force to (re-)download the model weights and configuration files and override the cached versions if they exists.
|
||||
|
||||
proxies: (`optional`) dict, default None:
|
||||
A dictionary of proxy servers to use by protocol or endpoint, e.g.: {'http': 'foo.bar:3128', 'http://hostname': 'foo.bar:4012'}.
|
||||
The proxies are used on each request.
|
||||
|
||||
output_loading_info: (`optional`) boolean:
|
||||
Set to ``True`` to also return a dictionnary containing missing keys, unexpected keys and error messages.
|
||||
|
||||
kwargs: (`optional`) Remaining dictionary of keyword arguments:
|
||||
Can be used to update the configuration object (after it being loaded) and initiate the model. (e.g. ``output_attention=True``). Behave differently depending on whether a `config` is provided or automatically loaded:
|
||||
|
||||
- If a configuration is provided with ``config``, ``**kwargs`` will be directly passed to the underlying model's ``__init__`` method (we assume all relevant updates to the configuration have already been done)
|
||||
- If a configuration is not provided, ``kwargs`` will be first passed to the configuration class initialization function (:func:`~transformers.PretrainedConfig.from_pretrained`). Each key of ``kwargs`` that corresponds to a configuration attribute will be used to override said attribute with the supplied ``kwargs`` value. Remaining keys that do not correspond to any configuration attribute will be passed to the underlying model's ``__init__`` function.
|
||||
|
||||
Examples::
|
||||
|
||||
model = AutoModel.from_pretrained('bert-base-uncased') # Download model and configuration from S3 and cache.
|
||||
model = AutoModel.from_pretrained('./test/bert_model/') # E.g. model was saved using `save_pretrained('./test/saved_model/')`
|
||||
model = AutoModel.from_pretrained('bert-base-uncased', output_attention=True) # Update configuration during loading
|
||||
assert model.config.output_attention == True
|
||||
# Loading from a TF checkpoint file instead of a PyTorch model (slower)
|
||||
config = AutoConfig.from_json_file('./tf_model/bert_tf_model_config.json')
|
||||
model = AutoModel.from_pretrained('./tf_model/bert_tf_checkpoint.ckpt.index', from_tf=True, config=config)
|
||||
|
||||
r""" Instantiates a pipeline from a pre-trained tokenizer and model.
|
||||
"""
|
||||
# Extract tokenizer and model arguments
|
||||
tokenizer_kwargs = {}
|
||||
@@ -159,9 +94,11 @@ class TextClassificationPipeline(object):
|
||||
# used for both the tokenizer and the model
|
||||
model_kwargs[key] = kwargs[key]
|
||||
|
||||
model_kwargs['output_loading_info'] = True
|
||||
tokenizer = AutoTokenizer.from_pretrained(pretrained_model_name_or_path, **tokenizer_kwargs)
|
||||
model = AutoModelForSequenceClassification.from_pretrained(pretrained_model_name_or_path, **model_kwargs)
|
||||
return cls(tokenizer, model)
|
||||
model, loading_info = AutoModelForSequenceClassification.from_pretrained(pretrained_model_name_or_path, **model_kwargs)
|
||||
|
||||
return cls(tokenizer, model, is_trained=bool(not loading_info['missing_keys']))
|
||||
|
||||
|
||||
def save_pretrained(self, save_directory):
|
||||
@@ -240,9 +177,13 @@ class TextClassificationPipeline(object):
|
||||
validation_data=valid_dataset, validation_steps=valid_steps, **kwargs)
|
||||
else:
|
||||
raise NotImplementedError
|
||||
self.is_trained = True
|
||||
|
||||
|
||||
def __call__(self, text):
|
||||
if not self.is_trained:
|
||||
logger.error("Some weights of the model are not trained. Please fine-tune the model on a classification task before using it.")
|
||||
|
||||
inputs = self.tokenizer.encode_plus(text, add_special_tokens=True, return_tensors=self.framework)
|
||||
if self.framework == 'tf':
|
||||
# TODO trace model
|
||||
|
||||
Reference in New Issue
Block a user