docs/UPLOAD_GUIDE.md

# Upload CHISÆL–NEXYTHRA II as a Hugging Face dataset

Suggested name: `CHISAEL-NEXYTHRA-II`.

Suggested display title: **CHISÆL–NEXYTHRA II: Nonlinear Helicity Event Research Dataset**.

Short description: **A 72-page AI-assisted mathematical draft and reproducibility dataset on finite nonlinear helicity-event calendars, quadratic retuning, phase-selected triad response, and energy-clock invariants. Includes 14 source tables, 1,139 verification records, code, figures, and claim provenance. Historical priority is unverified.**

## Browser upload

1. Extract this ZIP completely.
2. Open [Create a dataset](https://huggingface.co/new-dataset) and choose your account namespace and `CHISAEL-NEXYTHRA-II`. Set repository type to Dataset and select your desired visibility.
3. Upload the contents of the extracted `CHISAEL_NEXYTHRA_II_HF_Dataset_v1.0.0` folder, preserving subdirectories. `README.md`, `data/`, and `research/` belong directly at the repository root. A lone ZIP upload will not expose the structured viewer configurations.
4. Commit the files. Open the dataset page and inspect the default `verification_checks` configuration, then the other configurations. Viewer processing occurs on Hugging Face after upload; it has not been tested on a live repository by this preparation.

The source has no selected redistribution license. The card intentionally omits the YAML `license` field. If you decide on licensing, add the chosen terms and update the source notice/card rather than selecting a misleading Hub license tag.

## Python uploader (Windows, Linux, macOS)

Install the uploader dependency and authenticate once:

```bash
python -m pip install --upgrade huggingface_hub
hf auth login
```

Use a write-enabled token with access to your target dataset. The uploader reads saved authentication or `HF_TOKEN`; it does not ask you to paste a token into a file. Replace `YOUR_USERNAME` with your actual namespace:

```bash
python scripts/upload_dataset.py YOUR_USERNAME/CHISAEL-NEXYTHRA-II --dry-run
python scripts/upload_dataset.py YOUR_USERNAME/CHISAEL-NEXYTHRA-II
```

Use `--private` when creating a new private repository. The script does not change an existing repository's visibility. It validates release hashes, uploads only declared release files, and always uses `repo_type="dataset"`.

On Windows, double-click `upload_windows.bat` from the extracted folder; it asks for your repository ID and keeps the window open so you can read the result. PowerShell: `./upload.ps1 YOUR_USERNAME/CHISAEL-NEXYTHRA-II`. Linux/macOS: `bash upload.sh YOUR_USERNAME/CHISAEL-NEXYTHRA-II`.

If interrupted, rerun the same upload command and inspect the resulting dataset commit. No remote deletion is requested. If a file fails its release hash, extract a fresh copy or review your intended edits and regenerate the manifest deliberately.

## CLI alternative

From the extracted repository root, current documented CLI usage is:

```bash
hf upload YOUR_USERNAME/CHISAEL-NEXYTHRA-II . . --repo-type dataset
```

The Python uploader is the provided path that enforces the release manifest.

## Completion scope

This archive is prepared for upload; no repository was created or published during preparation. It contains no account tokens, secret credentials, invented DOI, or guessed account name. All 19 configurations were locally tested; live Hub viewer indexing still depends on the eventual upload.

Official references (checked 9 October 2026):

- [Dataset cards](https://huggingface.co/docs/hub/datasets-cards)
- [Data-file configuration](https://huggingface.co/docs/hub/datasets-data-files-configuration)
- [Manual configurations](https://huggingface.co/docs/hub/datasets-manual-configuration)
- [Upload API](https://huggingface.co/docs/huggingface_hub/guides/upload)
- [CLI](https://huggingface.co/docs/huggingface_hub/guides/cli)