/modern_chinese_nlp

(WIP) My humble contribution to the democratization of the Chinese NLP technology

Primary LanguageJupyter NotebookMIT LicenseMIT

Modern (Deep) Chinese NLP Models

My humble contribution to the democratization of the Chinese NLP technology (currently based on Fast.ai library).

[WIP] This project is still in a very early development stage. Things might change dramatically in the near future.

Blog post(s):

  1. [Preview] Developing Modern Chinese NLP Models - Briefly described dataset preparation processes and some preliminary results.

Environment

Pre-built docker image: docker pull ceshine/pytorch-fastai. Or build the image yourself with the accompanied Dockerfile.

Available models

Character-level Models

  • Language Model (Wikipedia Articles): next character prediction.
  • Sentiment Analysis (Douban Movie Reviews): movie rating prediction.

Word-level Models

  • Language Model (Wikipedia Articles): next character prediction.
  • Sentiment Analysis (Douban Movie Reviews): movie rating prediction.

Sub-Word-Level Models

  • Language Model (Wikipedia Articles): next character prediction.
  • Sentiment Analysis (Douban Movie Reviews): movie rating prediction.