ruichi_alpha

# ruichi_alpha Private dataset mirror of `~/work/ruichi_alpha` from v100. ## Layout This repo contains the folder split into multiple `.tar` archives for faster upload/download: - `_hf_min1_split/`: the original `ruichi_alpha/_hf_min1` directory (15 large yearly `.zip` files) packed into one `.tar` per zip. - `ruichi_alpha_data-base_misc.tar`: `ruichi_alpha/data-base` excluding the `warehouse` subdirectory (code, configs, logs, docs, state, etc.). - `ruichi_alpha_data-base_warehouse_*.tar`: small warehouse subdirectories (`daily`, `_meta`, `sectors`, ...). - `ruichi_alpha_data-base_warehouse_fin_ths_year=YYYY.tar`: `fin_ths` split by year. - `ruichi_alpha_data-base_warehouse_min1_year=YYYY.tar`: `min1` split by year. - `ruichi_alpha_data-base_warehouse_min5_year=YYYY.tar`: `min5` split by year. All archives preserve the `ruichi_alpha/...` relative path, so extracting them into the same destination will restore the original directory tree. ## Reconstruct locally ```bash # Download the whole dataset ms-hub download sunrich/ruichi_alpha --repo-type dataset --local-dir ./ruichi_alpha_ms # Or download with the ModelScope SDK from modelscope_hub import HubApi api = HubApi() api.download_repo("sunrich/ruichi_alpha", "dataset", local_dir="./ruichi_alpha_ms") # Extract all tars into a target directory mkdir -p ruichi_alpha cd ruichi_alpha for f in ../ruichi_alpha_ms/*.tar; do tar -xf "$f"; done for f in ../ruichi_alpha_ms/_hf_min1_split/*.tar; do tar -xf "$f"; done ``` After extraction you will have the full `ruichi_alpha/` tree as it was on v100.