ORI-Muchim's picture
Upload 8 files
d93648d
raw
history blame contribute delete
287 Bytes
# korean_cleaners
_pad = '_'
_punctuation = ',.!?…~'
_letters = 'γ„±γ„΄γ„·γ„Ήγ…γ…‚γ……γ…‡γ…ˆγ…Šγ…‹γ…Œγ…γ…Žγ„²γ„Έγ…ƒγ…†γ…‰γ…γ…“γ…—γ…œγ…‘γ…£γ…γ…” '
# Export all symbols:
symbols = [_pad] + list(_punctuation) + list(_letters)
# Special symbol ids
SPACE_ID = symbols.index(' ')