Datasetpublic

amitvpatel06/regime-bench

RegimeBench Base Data Bundle This dataset contains the base parquet inputs for the 18 public RegimeBench signal tasks. Split variants are generated locally from these base files by the benchmark repository rather than stored as a separate hosted payload. The data bundle is released under the Apache License 2.0, matching the benchmark source release. It contains processed benchmark data for research and reproducibility, not investment advice. Hidden/OOS labels are included, so… See the full description on the dataset page: https://huggingface.co/datasets/amitvpatel06/regime-bench.

sourceHugging Faceapache-2.0updated 4mo agoView on Hugging Face

0likes59downloads

Dataset Card

RegimeBench Base Data Bundle

This dataset contains the base parquet inputs for the 18 public RegimeBench signal tasks. Split variants are generated locally from these base files by the benchmark repository rather than stored as a separate hosted payload.

The data bundle is released under the Apache License 2.0, matching the benchmark source release. It contains processed benchmark data for research and reproducibility, not investment advice. Hidden/OOS labels are included, so this release supports local reproducibility rather than blind leaderboard evaluation.

Restore this bundle with:

bash

uv run python scripts/data_bundle.py hf-download --output-dir .release-data
uv run python scripts/data_bundle.py verify --bundle-dir .release-data
uv run python scripts/data_bundle.py restore --bundle-dir .release-data --force

Files are staged under payload/, with checksums in SHA256SUMS and bundle metadata in bundle_manifest.json.

Benchmark repository: https://github.com/tincans-ai/regime-bench