Skip to content

GitLab

Commit 035fea53 authored by Louis MARTIN's avatar Louis MARTIN Committed by Julien Chaumond
Browse files

Add CamemBERT to auto files and docs

parent 694d4fcb
Show whitespace changes
Inline Side-by-side
Showing with 18 additions and 4 deletions
docs/source/pretrained_models.rst
View file @ 035fea53
......@@ -155,5 +155,9 @@ Here is the full list of the currently provided pretrained models together with
| CTRL | ``ctrl`` | | 48-layer, 1280-hidden, 16-heads, 1.6B parameters |
| | | | Salesforce's Large-sized CTRL English model |
+-------------------+------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------+
| CamemBERT | ``camembert-base`` | | 12-layer, 768-hidden, 12-heads, 110M parameters |
| | | | CamemBERT using the BERT-base architecture |
| | | (see `details <https://github.com/pytorch/fairseq/tree/master/examples/camembert>`__) |
+-------------------+------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------------------+
.. <https://huggingface.co/transformers/examples.html>`__
transformers/configuration_auto.py
View file @ 035fea53
......@@ -27,6 +27,7 @@ from .configuration_xlm import XLMConfig
from .configuration_roberta import RobertaConfig
from .configuration_distilbert import DistilBertConfig
from .configuration_ctrl import CTRLConfig
from .configuration_camembert import CamembertConfig
logger = logging.getLogger(__name__)
......@@ -50,6 +51,7 @@ class AutoConfig(object):
- contains `xlnet`: XLNetConfig (XLNet model)
- contains `xlm`: XLMConfig (XLM model)
- contains `roberta`: RobertaConfig (RoBERTa model)
- contains `camembert`: CamembertConfig (CamemBERT model)
- contains `ctrl` : CTRLConfig (CTRL model)
This class cannot be instantiated using `__init__()` (throw an error).
"""
......@@ -72,6 +74,7 @@ class AutoConfig(object):
- contains `xlnet`: XLNetConfig (XLNet model)
- contains `xlm`: XLMConfig (XLM model)
- contains `roberta`: RobertaConfig (RoBERTa model)
- contains `camembert`: CamembertConfig (CamemBERT model)
- contains `ctrl` : CTRLConfig (CTRL model)
Params:
pretrained_model_name_or_path: either:
......@@ -116,6 +119,8 @@ class AutoConfig(object):
"""
if 'distilbert' in pretrained_model_name_or_path:
return DistilBertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
elif 'camembert' in pretrained_model_name_or_path:
return CamembertConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
elif 'roberta' in pretrained_model_name_or_path:
return RobertaConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
elif 'bert' in pretrained_model_name_or_path:
......@@ -134,4 +139,4 @@ class AutoConfig(object):
return CTRLConfig.from_pretrained(pretrained_model_name_or_path, **kwargs)
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
"'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
"'xlm', 'roberta', 'ctrl'".format(pretrained_model_name_or_path))
"'xlm', 'roberta', 'camembert', 'ctrl'".format(pretrained_model_name_or_path))
transformers/tokenization_auto.py
View file @ 035fea53
......@@ -27,6 +27,7 @@ from .tokenization_xlnet import XLNetTokenizer
from .tokenization_xlm import XLMTokenizer
from .tokenization_roberta import RobertaTokenizer
from .tokenization_distilbert import DistilBertTokenizer
from .tokenization_camembert import CamembertTokenizer
logger = logging.getLogger(__name__)
......@@ -41,6 +42,7 @@ class AutoTokenizer(object):
The tokenizer class to instantiate is selected as the first pattern matching
in the `pretrained_model_name_or_path` string (in the following order):
- contains `camembert`: CamembertTokenizer (CamemBERT model)
- contains `distilbert`: DistilBertTokenizer (DistilBert model)
- contains `roberta`: RobertaTokenizer (RoBERTa model)
- contains `bert`: BertTokenizer (Bert model)
......@@ -64,8 +66,9 @@ class AutoTokenizer(object):
The tokenizer class to instantiate is selected as the first pattern matching
in the `pretrained_model_name_or_path` string (in the following order):
- contains `camembert`: CamembertTokenizer (CamemBERT model)
- contains `distilbert`: DistilBertTokenizer (DistilBert model)
- contains `roberta`: RobertaTokenizer (XLM model)
- contains `roberta`: RobertaTokenizer (RoBERTa model)
- contains `bert`: BertTokenizer (Bert model)
- contains `openai-gpt`: OpenAIGPTTokenizer (OpenAI GPT model)
- contains `gpt2`: GPT2Tokenizer (OpenAI GPT-2 model)
......@@ -103,6 +106,8 @@ class AutoTokenizer(object):
"""
if 'distilbert' in pretrained_model_name_or_path:
return DistilBertTokenizer.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 'roberta' in pretrained_model_name_or_path:
return RobertaTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
elif 'bert' in pretrained_model_name_or_path:
......@@ -121,4 +126,4 @@ class AutoTokenizer(object):
return CTRLTokenizer.from_pretrained(pretrained_model_name_or_path, *inputs, **kwargs)
raise ValueError("Unrecognized model identifier in {}. Should contains one of "
"'bert', 'openai-gpt', 'gpt2', 'transfo-xl', 'xlnet', "
"'xlm', 'roberta', 'ctrl'".format(pretrained_model_name_or_path))
"'xlm', 'roberta', 'camembert', 'ctrl'".format(pretrained_model_name_or_path))
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment