UPLOAD_GUIDE.md

# Upload this release to Hugging Face

Suggested repository slug: **AILITH-Beyond-Reinforcement-Learning**  
Repository type: **Dataset**, used here to host research artifacts.  
Display name: **AÍLITH — Beyond Reinforcement Learning**

The kit is prepared for upload. No Hugging Face repository has been created or changed by its preparation.

## Windows uploader

1. Extract the complete release ZIP to a normal folder.
2. Open `AILITH_HuggingFace_v0.5.0` and double-click **`UPLOAD_TO_HF.bat`**.
3. Follow the standard Hugging Face login if needed. Enter the repository name or paste its dataset URL.
4. Choose visibility for a new repository, review the displayed destination, and start the upload.

The launcher prepares an isolated Python environment and installs the Hub client when needed. Python 3.10 or newer and internet access are required. If Python is missing, the launcher explains how to install it.

The uploader checks `SHA256SUMS` before upload and selects only the listed release files plus that checksum file. Repository contents are uploaded at the root so the card and relative links work. Other remote files are kept, and an existing repository retains its visibility. Transient failures can be retried; an interrupted upload can be rerun to the same destination.

See [UPLOADER_README.md](UPLOADER_README.md) for offline preview, command-line options, and recovery instructions. The uploader-only download also works with the earlier release folder or its outer ZIP.

## Browser upload

1. Create a dataset repository under your chosen account or organization. Choose its visibility in the creation form.
2. Extract the outer `AILITH_HuggingFace_Release_v0.5.0.zip`.
3. Open its `AILITH_HuggingFace_v0.5.0` directory.
4. Upload the **contents** of that directory into the repository root. `README.md` must be at the root, alongside `LICENSE`, citation files, and the artifact directories.
5. Commit, then check that the card renders and that its PDF and full-archive links work.

Uploading only the outer ZIP will leave the root card and its links unavailable to Hub rendering. The original research ZIP under `releases/` should remain a ZIP; it is the downloadable reproducibility package.

## CLI upload

The current Hugging Face CLI is `hf`. Install the Hub package and authenticate using its standard interactive login:

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

From the directory containing `AILITH_HuggingFace_v0.5.0`, replace `YOUR_NAMESPACE` with your actual account or organization:

```bash
hf repos create YOUR_NAMESPACE/AILITH-Beyond-Reinforcement-Learning --repo-type dataset
hf upload YOUR_NAMESPACE/AILITH-Beyond-Reinforcement-Learning ./AILITH_HuggingFace_v0.5.0 . --repo-type dataset --commit-message "AÍLITH v0.5.0: complete research release"
```

Add `--private` to repository creation if that is the visibility you want. For an existing repository, skip creation and upload to the selected repository ID. The commands do not request file deletion or alter visibility of an existing repository. Obtain exact local option details with `hf repos create --help` and `hf upload --help` if your installed CLI differs.

## Description

Copy `REPOSITORY_DESCRIPTION.txt` into a short-description field where available. The detailed publication text is already in `README.md`; its YAML declares English, the existing MIT license, and relevant research tags. The manuscript title and requested author remain unchanged.

## Metadata and citation

This is an artifact release rather than a pretrained model or conventional row-based training dataset. A dataset preview is not required for reading the manuscript or running the software. No model task tag, trained weights, DOI, or arXiv identifier is supplied. Add an actual repository URL or DOI to the citation files only after one exists.

The official references used for this upload layout are the [dataset-card documentation](https://huggingface.co/docs/hub/datasets-cards) and [Hub upload guide](https://huggingface.co/docs/huggingface_hub/guides/upload), checked on 18 October 2026.