Embedding/Chinese-Word-Vectors

Provides both dense (SGNS) and sparse (PPMI) embedding representations across diverse corpora (encyclopedias, news, social media, classical literature) with flexible context features (word, n-gram, character combinations) to capture different morphological and semantic properties. Includes the CA8 Chinese analogical reasoning benchmark and evaluation toolkit for intrinsic quality assessment, enabling users to benchmark vector performance on downstream NLP tasks.

12,188 stars. No commits in the last 6 months.