Skip to content

Latest commit

 

History

History
49 lines (38 loc) · 1.66 KB

changelog.md

File metadata and controls

49 lines (38 loc) · 1.66 KB
Raw

monpa v0.3.3

  • Upgrade torch API w.r.t. where

monpa v0.3.2

  • fix issue 10, 11. related to load_userdict and logging.
  • Add "utils.short_sentence", "utils.cut_mp" and "utils.pseg_mp" method for shorting long sentence and implement multiprocessing to cut and pseg function.

monpa v0.3.1

  • Add "cut_batch" and "pseg_batch" method to support batch segmentation by using GPU computing.

monpa v0.3.0.0

  • change model structure to reduce file size and increase speed
  • no need to download model file
  • enable userdict weight-value setting

monpa v0.2.6

monpa v0.2.5.1

  • force incomplete sentence can be output word and pos without error message. ex.『隔離百分之』can be output as [['隔離', 'VC'], ['百分之', 'Neqa']], although "百分之" is incomplete.
  • fix load_userdict function to avoid output dubplicate user define terms.
  • fintune some wording.

monpa v0.2.5

  • fine turn LICENSE to BSD
  • revision __init__.py and README.md
  • revise author email at setup.py
  • release date: 2019/07/26

monpa v0.2.4.1

  • add package dependence to check torch >= 1.0 and requests, if false, pip install directly.

monpa v0.2.4

  • fix model folder issue

monpa v0.2.3

  • due to pypi file size limitation, revision for not include model file in wheel but invoking download when 1st time import.

monpa v.0.2.2

  • add cut, pseg, load_userdict function
  • fix file path issue
  • fix CP950 issue: when open file by setting encoding="utf-8"
  • setup.py revision
  • README.md revision
  • release date: 2019/07/23

monpa v0.2.1

  • adopt new training model