Tencent AI Lab Embedding Corpus

One of the most widely used datasets for Chinese NLP, providing 200-dimension vectors for 8 million Chinese words and phrases.