Datasets:

PureOne
/

Infinite-Players

No account namespace has been assumed. Replace YOUR_USERNAME below with the actual account or organization that will own the repository. These instructions prepare publication; no remote repository has been created or modified by this release-preparation task.

The archive supplies the root README.md, metadata, paper, editable sources, code, experiments, result tables, and documentation. Upload its extracted contents to the repository root. Uploading only the ZIP would hide the detailed card and relative links from the repository landing page.

Windows uploader (included)

Extract the full release ZIP and double-click UPLOAD_TO_HUGGINGFACE.bat. The launcher creates a local .hf-uploader-venv and installs only the upload requirements. It then runs UPLOAD_TO_HUGGINGFACE.py from the extracted folder. Python 3.11 or newer and an internet connection are needed for the launcher.

The uploader checks release hashes, uses your cached Hugging Face login or HF_TOKEN, and otherwise asks for a token with hidden input. A token entered at this prompt is used for the current run and is not written into the package. Use a token with write access to the destination dataset repository. You can create one at Hugging Face token settings.

Choose the suggested YOUR_USERNAME/Infinite-Players destination or enter an organization repository. The uploader shows the account, destination, exact file count, and visibility before requesting upload confirmation. A new repository defaults to public; choose private when prompted if preferred. Existing repository visibility is preserved. Uploading updates matching files and does not delete other remote files.

Only files listed in FILE_MANIFEST.json, plus that manifest itself, enter a verified temporary staging folder. Extra local files, tokens, virtual environments, and generated outputs are excluded. A changed listed file causes the upload to stop with a clear diagnostic; use the intact release or rebuild its manifest deliberately after making your own changes.

Portable Python uploader

From the extracted repository root:

python -m pip install -r requirements-upload.txt
python UPLOAD_TO_HUGGINGFACE.py

To specify a repository explicitly:

python UPLOAD_TO_HUGGINGFACE.py --repo-id YOUR_USERNAME/Infinite-Players

To create a new private repository:

python UPLOAD_TO_HUGGINGFACE.py --repo-id YOUR_USERNAME/Infinite-Players --private

For an offline preview that needs neither an account nor upload dependencies:

python UPLOAD_TO_HUGGINGFACE.py --dry-run

--dry-run checks the manifest and prints the upload plan without network calls. The optional --yes switch skips the final interactive upload confirmation for an explicitly supplied repository. Run the same command again after a network failure to update the same repository. No remote upload was run while preparing this release; uploader behavior is tested with a fake Hub client.

Browser upload

Extract the release ZIP. Open the extracted Infinite-Players-HuggingFace directory; README.md and REPRODUCE.py should be directly inside it.
In Hugging Face, choose New Dataset and enter the namespace and suggested repository name. Set the visibility you intend.
Preserve the current no-public-license status. Do not select a standard license unless you intend to grant it and add matching license terms. See LICENSE_STATUS.txt.
In Files and versions, upload the extracted files and subdirectories, preserving their relative paths. Place README.md at the repository root.
Use a commit message such as Release Infinite Players v3.0.0 research edition.
Open the main card and check the manuscript link, the two inline figures, technical overview, and validation links. viewer: false is intentional: these are heterogeneous research artifacts, not a single tabular dataset.

CLI upload

Install the current Hugging Face CLI if needed. From the extracted repository root, verify the local release before uploading:

python VERIFY_RELEASE.py
hf auth login
hf repos create YOUR_USERNAME/Infinite-Players --repo-type dataset --exist-ok
hf upload YOUR_USERNAME/Infinite-Players . . --repo-type dataset --commit-message "Release Infinite Players v3.0.0 research edition" --exclude "__pycache__/*" "*.pyc" "demo_output/*" "reproduced*/*" ".venv/*" ".cache/*" "*.egg-info/*"

This command creates a public repository by default if it does not already exist. If you intend a private first upload, add --private to the create command. --exist-ok does not change an existing repository's visibility. Use a new repository for this curated edition if an old repository contains unrelated files: this upload does not delete existing remote files.

The upload command excludes common generated directories but is best run from a fresh extraction. A Hugging Face .gitignore is not a substitute for checking what a general folder upload includes. Never put an access token in the source, README, command history, or an uploaded configuration file. hf auth login handles local authentication; the supplied research itself needs no account.

The included Python uploader uses the huggingface_hub API. The manual CLI commands above are an alternative. Use hf upload --help for the installed CLI version; current documentation also accepts --type as an alias for --repo-type.

Short description

Use the following if the interface asks for a short description:

Finite mathematical research and runnable Python on semantic-trajectory-preserving game-agent improvement, revisable temporal evidence, future-calibration memory, and auditable synthetic benchmarks.

The extended release description is in docs/RELEASE_DESCRIPTION.md. The root README is the complete landing-page card. Neither claims trained model weights, real-player reconstruction, or completed commercial-engine support.

After publication

Add the real repository URL and a pinned revision to citations or release announcements. No DOI, arXiv identifier, external peer review, or model endpoint is created by uploading this archive. If you later choose a public license, include its exact terms and update the card metadata in the same revision.

Official platform references

Platform instructions were checked against the official documentation during preparation on 18 October 2026:

These references document the hosting workflow, not the scientific claims.