dnrops
/
transformers
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

clean up in tokenization

This commit is contained in:
thomwolf 2019-05-08 21:43:51 +02:00
parent 0efc4ab632
commit 366a3b0285
3 changed files with 13 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
pytorch_pretrained_bert/modeling_gpt2.py
View File

@ -39,8 +39,10 @@ from .modeling import BertLayerNorm as LayerNorm
logger = logging.getLogger(__name__)
PRETRAINED_MODEL_ARCHIVE_MAP = {"gpt2": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-pytorch_model.bin"}
PRETRAINED_CONFIG_ARCHIVE_MAP = {"gpt2": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-config.json"}
PRETRAINED_MODEL_ARCHIVE_MAP = {"gpt2": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-pytorch_model.bin",
"gpt2-medium": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-medium-pytorch_model.bin"}
PRETRAINED_CONFIG_ARCHIVE_MAP = {"gpt2": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-config.json",
"gpt2-medium": "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-medium-config.json"}
def load_tf_weights_in_gpt2(model, gpt2_checkpoint_path):
""" Load tf checkpoints in a pytorch model

9
pytorch_pretrained_bert/tokenization_gpt2.py
View File

@ -37,9 +37,11 @@ logger = logging.getLogger(__name__)
PRETRAINED_VOCAB_ARCHIVE_MAP = {
'gpt2': "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-vocab.json",
'gpt2-medium': "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-medium-vocab.json",
}
PRETRAINED_MERGES_ARCHIVE_MAP = {
'gpt2': "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-merges.txt",
'gpt2-medium': "https://s3.amazonaws.com/models.huggingface.co/bert/gpt2-medium-merges.txt",
}
PRETRAINED_VOCAB_POSITIONAL_EMBEDDINGS_SIZE_MAP = {
'gpt2': 1024,
@ -263,9 +265,14 @@ class GPT2Tokenizer(object):
def encode(self, text):
return self.convert_tokens_to_ids(self.tokenize(text))
def decode(self, tokens, skip_special_tokens=False):
def decode(self, tokens, skip_special_tokens=False, clean_up_tokenization_spaces=True):
text = ''.join(self.convert_ids_to_tokens(tokens, skip_special_tokens=skip_special_tokens))
text = bytearray([self.byte_decoder[c] for c in text]).decode('utf-8', errors=self.errors)
if clean_up_tokenization_spaces:
text = text.replace('<unk>', '')
text = text.replace(' .', '.').replace(' ?', '?').replace(' !', '!').replace(' ,', ','
).replace(" ' ", "'").replace(" n't", "n't").replace(" 'm", "'m").replace(" do not", " don't"
).replace(" 's", "'s").replace(" 've", "'ve").replace(" 're", "'re")
return text
def save_vocabulary(self, vocab_path):

2
pytorch_pretrained_bert/tokenization_openai.py
View File

@ -272,7 +272,7 @@ class OpenAIGPTTokenizer(object):
out_string = ''.join(tokens).replace('</w>', ' ').strip()
if clean_up_tokenization_spaces:
out_string = out_string.replace('<unk>', '')
out_string = out_string.replace(' .', '.').replace(' ?', '?').replace(' !', '!').replace(' ,', ',').replace(' ,', ','
out_string = out_string.replace(' .', '.').replace(' ?', '?').replace(' !', '!').replace(' ,', ','
).replace(" ' ", "'").replace(" n't", "n't").replace(" 'm", "'m").replace(" do not", " don't"
).replace(" 's", "'s").replace(" 've", "'ve").replace(" 're", "'re")
return out_string