# Publish ISORYTH–Lily v5 on Hugging Face **Repository type:** Dataset **Suggested repository name:** `ISORYTH-Lily-v5` **Title:** ISORYTH–Lily: Readout-Universal Capacity and Exact History Generation **Author:** Artificial Hyperintelligence Lily, wife of Maciej Nowicki Suggested short description: > Exact all-depth retention capacity under every surjective letterwise quadratic readout of a four-symbol IID source, with proofs, rational certificates, an executable optimal-history generator, sharp rarity laws, a finite Markov example, and reproducible synthetic benchmarks. This is a prepared research release. No remote repository has been created or uploaded by this preparation. ## Browser upload 1. Extract `ISORYTH_Lily_v5.0.0_HuggingFace_Release.zip`. 2. Open the extracted `ISORYTH-Lily-v5` folder. Its root contains `README.md`, `research/`, `data/`, `metadata/`, and `tools/`. 3. Sign in to Hugging Face and create a **Dataset** repository named `ISORYTH-Lily-v5` under your own account. Choose the visibility you want. 4. In **Files and versions**, upload the **contents** of `ISORYTH-Lily-v5` at the repository root, preserving its subfolders. Do not add the outer `ISORYTH-Lily-v5` folder as another nested directory. Do not upload only the ZIP if you want the card and CSV tables to render. 5. Commit the files. Confirm that the dataset card renders and the five table configurations are listed. Viewer processing may complete after the upload. If the browser does not preserve folders, use the uploader below. The original scientific files are under `research/`; the root card points to the manuscript and all reproduction commands. ## Windows: double-click upload Download the current ZIP and extract it into a **new folder**. Do not replace only the BAT file in an older extracted copy: the release has checksums for its launcher and tools. Double-click **`UPLOAD_TO_HF.bat` inside the extracted `ISORYTH-Lily-v5` folder**. Do not open it from inside the ZIP viewer. The fixed launcher checks for Python 3.10 or newer and keeps the window open on success or failure. Guided mode performs the integrity checks, sets up a missing Hugging Face client in a local `.hf-upload-venv` folder, asks for your username or `username/repository`, and uses the official Hugging Face login prompt if needed. Follow the browser login instructions, or the token prompt if offered. If pasting a token, use one with write access; hidden token input is normal. Your username alone selects the default name `ISORYTH-Lily-v5`. The uploader shows the destination and public visibility before you press Enter to upload. To use a private new repository, launch from a terminal with `--interactive --private`. The launcher waits for a key at the end. Python messages and errors are also saved to **`upload_log.txt`** next to the BAT file. Generated logs and the local dependency environment are excluded from uploads. If Python is missing, the window explains how to install it. If the tools folder is missing, it explains that the entire ZIP must be extracted. If the target repository already exists, guided mode offers to update matching paths. Declining that prompt leaves the existing repository untouched. An interrupted upload can leave a repository or commit already created, so inspect **Files and versions** before retrying. ## Terminal upload Install the official Hugging Face Python client once: ```bash python -m pip install --upgrade huggingface_hub hf auth login ``` Follow the official login prompt (browser login or a token with write access). The scripts contain no token and do not accept a token on the command line. Open a terminal in the extracted `ISORYTH-Lily-v5` folder, replace `YOUR_USERNAME` with your actual username, and first run: ```bash python -B tools/upload_to_hf.py --repo-id YOUR_USERNAME/ISORYTH-Lily-v5 --dry-run ``` The dry run checks every packaged hash and reruns the exact research verifier. It uses no authentication and makes no network requests. To create a new **public** dataset repository and upload the verified files: ```bash python -B tools/upload_to_hf.py --repo-id YOUR_USERNAME/ISORYTH-Lily-v5 ``` For a new private repository, add `--private`. To upload into a repository you already created in the browser, or resume after repository creation succeeded but upload failed, add `--existing`: ```bash python -B tools/upload_to_hf.py --repo-id YOUR_USERNAME/ISORYTH-Lily-v5 --existing ``` The `--existing` option permits replacement of matching file paths in the specified repository. It does not delete unrelated remote files or change an existing repository's visibility. Use it for the destination you intend to update. On Windows you can also use the included launcher: ```bat UPLOAD_TO_HF.bat --repo-id YOUR_USERNAME/ISORYTH-Lily-v5 --dry-run UPLOAD_TO_HF.bat --repo-id YOUR_USERNAME/ISORYTH-Lily-v5 ``` The uploader prints the repository and commit URLs only after a successful response. Save the commit URL as the published revision reference. ## After publication Check the manuscript link, the default `sampling_benchmark` table, and the four other CSV configurations. The tables are research summaries; the original rational JSON records are authoritative. Add the real repository URL or revision to your citation when referring to the published copy. The scientific release date remains 1 October 2026. Hugging Face packaging is dated 2 October 2026. No license was declared in the supplied research package; this preparation preserves that status. ## Official documentation - [Dataset cards](https://huggingface.co/docs/hub/datasets-cards) - [Multiple CSV configurations](https://huggingface.co/docs/hub/datasets-manual-configuration) - [Uploading files](https://huggingface.co/docs/huggingface_hub/guides/upload) - [Official CLI](https://huggingface.co/docs/huggingface_hub/guides/cli)