Asset & Text Embeddings for the China A-share Market

Asset and text embeddings for the China A-share market, accompanying "Time-varying Asset Embeddings in China" (Zitong Cheng, Dashan Huang, Xiaobin Liu, Zeng Tao). asset_embeddings.tar.gz — learned per-stock embeddings (AssetBERT, AssetW2V, AssetRS) at dimensions 4–128 for six quarters (2023Q2–2024Q3). text_embeddings.tar.gz — a compact, fully reproducible core of nine multilingual text-embedding baselines over A-share company names: full-dimensional masters plus regenerate.py, which rebuilds any transform × dimension × quarter variant. See README.md for usage, schema, and the reproducibility note. Company names derive from CSMAR.