---
configs:
  - config_name: sample_50
    data_files:
      - split: train
        path: "sample_50/metadata.csv"
  - config_name: preview_1000
    data_files:
      - split: train
        path: "preview_1000/metadata.csv"

license: other
license_name: zion-vintage-media-proprietary
license_link: LICENSE
task_categories:
- text-generation
- image-to-text
- visual-question-answering
language:
- en
tags:
- historical
- recipes
- multimodal
- ocr
- alternative-data
- Text-Generation
- Historical
size_categories:
- 1K<n<10K
pretty_name: Zion Vintage Recipes (Multimodal Archive)
---

# Zion Vintage Media: Historical Multimodal Culinary Archive

## 💎 Commercial Licensing & Full Dataset Access
**This repository contains two evaluation tiers.** The full, production-ready dataset contains **27,000+ multimodal pairs** (High-Resolution Scans + Structured JSON) of unpublished, offline historical recipes spanning from the 1940s to the 2010s. 

This data is **Copyright Cleared** and is **Internet-Original**, meaning it has never been indexed by web crawlers, making it perfect for training models on "uncontaminated" tokens.

To purchase a commercial license for the full SQL database and high-resolution TIF/JPG scans, please contact:
* **Email:** admin@zionarchive.ai
* **Website:** [Zion Archive](https://www.zionarchive.ai/)
* **Live Showcase:** [Interactive Dataset Explorer](https://recipes.zionarchive.ai/)

---

## 📂 Available Versions
Use the Hugging Face `config` selector to load your preferred scale:

| Version | Size | Purpose |
| :--- | :--- | :--- |
| **sample_50** | 50 Rows | Instant format check and schema validation. |
| **preview_1000** | 1,000 Rows | Research preview for loss-curve testing & mini-finetuning. |

### How to load the 1,000-Row Preview:
```python
from datasets import load_dataset

dataset = load_dataset("dataDatasic/historical-culinary-archive-multimodal", "preview_1000")