Releases: Yoctol/uttut
Releases · Yoctol/uttut
1.4.10: New Operators (AddWhitespaceAroundWordnZhChar, ReplaceAwithB)
1.4.9: Add Operator - PureNum2Token
Main modifications:
- #132 Added operator - PureNum2Token
1.4.8: [Fix BUG] Forgot setting `include_package_data = True`
Main Modifications:
- #131 Set
include_package_data = True
insetup.py
, so non-code files can be added
in the package when installing.
1.4.7: [Fix BUG] Forget uploading stopwords.txt
Main Modifications:
- #130 Fixed BUG of version 1.4.6.
We forgot uploading stopwords.txt to PYPI.
1.4.6: More Operations
Main modifications:
- (#123): Create an operation which can replace chinese stopwords with whitespace.
- (#124): Developers can indicate a specific file or folder when running unit tests.
- (#125): Operations which replace digits with special tokens now can handle fullwidth digits.
- (#126): Create an operation - NumTokenwithSpace which can replace continuous digits with token
_num_
.
Note that it is 87% similar to IntTokenwithSpace. - (#127): Let everyone know that we can reproduce Bert preprocessing through pipe.
- (#128): Users can check version of
uttut
via the following commands:
>> import uttut
>> uttut.__version__
1.4.6
1.4.5: Human Readable Dump
1.4.4: Pipe utility
Pipe supports +, +=, getitem, copy
1.4.3: Pipe Summary
support pipe.summary() to show detail steps & checkpoints
1.4.2: Import Sugar
can import Pipe from uttut.pipeline
1.4.1: Backward Compatibility
Let Pipe.deserialize support old format