[BUG] Docstring typo and missing at transformers.sequence_classification.Processor.text_pair_transform
loomlike opened this issue · comments
Jun Ki Min commented
Description
Since the issues are tiny, I put two issues at one github-issue:
- Typo at
transformers.sequence_classification.Processor.text_pair_transform
docstring,
@staticmethod
def text_pair_transform(text_1, text_2, tokenizer, max_len=MAX_SEQ_LEN):
"""
Text transformation function for sentence pair classification.
The function can be passed to a map-style PyTorch DataSet.
Args:
text_1 (str): Input text 1.
text_1 (str): Input text 2.
second parameter should be text_2
create_dataloader_from_df
doesn't have a docstring. Would be nice to have the docstring about argument and how to use the function since the function generates either single-sequence classification or sequence-pair classification based on the arguments.
Expected behavior (i.e. solution)
- Correct parameter name at
text_pair_transform
- Docstring at
create_dataloader_from_df