Upload README.md with huggingface_hub

README.md

@@ -43,7 +43,7 @@ configs:
 [![Total Articles](https://img.shields.io/badge/Articles-680%20Posts-blueviolet?style=flat-square)](https://huggingface.co/datasets/sphita/openai-news)
 [![GitHub Org](https://img.shields.io/badge/GitHub-sphita-181717?style=flat-square&logo=github&logoColor=white)](https://github.com/sphita)
 
-[Overview](#-overview--executive-summary) · [Corpus & Token Stats](#-comprehensive-corpus--token-statistics) · [Section Distribution](#-section--topic-distribution) · [Yearly Trends](#-historical-yearly-distribution) · [Schema](#-dataset-schema) · [Quick Start](#-quick-start-examples)
+[Overview](#-overview--executive-summary) · [Corpus & Token Stats](#-comprehensive-corpus--token-statistics) · [Multimodal Assets](#-multimodal-media--visual-assets-archive) · [Section Distribution](#-section--topic-distribution) · [Yearly Trends](#-historical-yearly-distribution) · [Schema](#-dataset-schema) · [Quick Start](#-quick-start-examples)
 
 </div>
 
@@ -59,7 +59,7 @@ Spanning **11+ years of transformative AI history** (2015-12-11 to 2026-10-15),
 * **Zero Container Stutter**: Custom parsing eliminates navigation noise, sidebar TOC anchors, and list container stutter.
 * **Exact Token Auditing**: Profiled using OpenAI's standard `cl100k_base` BPE tokenizer, measuring **1,095,798 total tokens**.
 * **Zero-Leak Parquet & Streaming JSONL**: Distributed in Zstandard-compressed Parquet (`train-00000-of-00001.parquet`) and streaming JSON Lines (`openai_news.jsonl`).
-* **Complete Multimodal Assets**: Tracks high-resolution cover hero images, inline diagrams, and full-page high-resolution screenshots.
+* **Complete Multimodal Assets**: Bundled with all 3,200 high-resolution cover hero images and inline illustrations (`images.tar.gz`), plus 680 full-page high-resolution screenshots (`screenshots.tar.gz`).
 
 ---
 
@@ -75,22 +75,46 @@ Spanning **11+ years of transformative AI history** (2015-12-11 to 2026-10-15),
 | **Average Tokens per Article** | **1611.5** tokens | — | **1611.5** tokens |
 | **Median Tokens per Article** | **1,267** tokens | — | **1,267** tokens |
 | **Parquet Size (Zstandard-9)** | — | — | **2.98 MB** |
+| **JSONL Streaming Size** | — | — | **11.79 MB** |
 | **Columnar Compression Ratio** | — | — | **1.7x reduction** |
 
 ---
 
+## 🖼 Multimodal Media & Visual Assets Archive
+
+All media referenced across the articles are packaged into compressed archives for clean distribution:
+
+| Asset Archive | Description | File Count | Archive Size | Hugging Face Repository Path |
+| :--- | :--- | :--- | :--- | :--- |
+| **`images.tar.gz`** | All featured cover images and inline architectural diagrams | 3,200 files | **1057.2 MB** | [`media/images.tar.gz`](https://huggingface.co/datasets/sphita/openai-news/resolve/main/media/images.tar.gz) |
+| **`screenshots.tar.gz`** | Pixel-perfect, cookie-free full-render screenshots of all articles | 680 files | **298.9 MB** | [`media/screenshots.tar.gz`](https://huggingface.co/datasets/sphita/openai-news/resolve/main/media/screenshots.tar.gz) |
+
+### Unpacking Media Assets
+
+```bash
+# Download and unpack images
+huggingface-cli download sphita/openai-news media/images.tar.gz --local-dir ./
+tar -xzf media/images.tar.gz
+
+# Download and unpack full-page screenshots
+huggingface-cli download sphita/openai-news media/screenshots.tar.gz --local-dir ./
+tar -xzf media/screenshots.tar.gz
+```
+
+---
+
 ## 🗂️ Section & Topic Distribution
 
 | Section Name | Article Count | Clean Words | cl100k Tokens | % of Corpus | Subject Matter & Scope |
 | :--- | :--- | :--- | :--- | :--- | :--- |
-| **Company** | 217 | 176,004 | 225,679 | 31.9% | Corporate governance, philanthropic partnerships, educational initiatives, infrastructure milestones |
+| **Company** | 229 | 199,014 | 256,206 | 33.7% | Corporate governance, philanthropic partnerships, educational initiatives, infrastructure milestones |
 | **Product** | 181 | 241,205 | 330,417 | 26.6% | Public foundation model releases, ChatGPT capabilities, API platform tooling, and enterprise features |
 | **Safety** | 147 | 178,404 | 226,742 | 21.6% | Frontier AI safety evaluations, red teaming, preparedness framework, watermarking, and cyber defense |
-| **Research** | 54 | 102,889 | 134,218 | 7.9% | Fundamental deep learning theory, RL algorithms, reasoning, alignment, and model interpretability |
+| **Research** | 42 | 79,879 | 103,691 | 6.2% | Fundamental deep learning theory, RL algorithms, reasoning, alignment, and model interpretability |
 | **Security** | 34 | 45,779 | 57,672 | 5.0% | Threat intelligence, counter-espionage investigations, disruption of malicious influence campaigns |
 | **Engineering** | 20 | 62,246 | 83,108 | 2.9% | Large-scale infrastructure engineering, distributed supercomputing clusters, inference optimization |
-| **Global Affairs** | 16 | 16,411 | 20,404 | 2.4% | Frontier technical documentation and releases |
-| **AI Adoption** | 9 | 9,364 | 11,848 | 1.3% | Frontier technical documentation and releases |
+| **Global Affairs** | 16 | 16,411 | 20,404 | 2.4% | International policy, multilateral governance, educational equity, economic impact studies |
+| **AI Adoption** | 9 | 9,364 | 11,848 | 1.3% | Real-world institutional deployments across scientific labs, enterprise systems, and classrooms |
 | **new platform** | 1 | 2,951 | 3,507 | 0.1% | Frontier technical documentation and releases |
 | **this blog** | 1 | 1,839 | 2,203 | 0.1% | Frontier technical documentation and releases |
 
@@ -127,10 +151,10 @@ Spanning **11+ years of transformative AI history** (2015-12-11 to 2026-10-15),
 | `body_text` | `string` | Normalized plain text content |
 | `body_markdown` | `string` | Structured markdown with headings, code blocks, lists |
 | `featured_image` | `string` | Remote CDN URL for high-resolution cover image |
-| `featured_image_path` | `string` | Local relative filesystem path to downloaded cover image |
+| `featured_image_path` | `string` | Local relative filesystem path to downloaded cover image (`images/<slug>/featured.*`) |
 | `inline_images` | `list<string>` | Remote CDN URLs for inline architectural illustrations |
-| `inline_image_paths` | `list<string>` | Local relative filesystem paths to downloaded inline images |
-| `screenshot_path` | `string` | Local relative filesystem path to full-page render screenshot |
+| `inline_image_paths` | `list<string>` | Local relative filesystem paths to downloaded inline images (`images/<slug>/inline_*.*`) |
+| `screenshot_path` | `string` | Local relative filesystem path to full-page render screenshot (`screenshots/<slug>.png`) |
 
 ---
 
@@ -148,6 +172,8 @@ for article in dataset:
     print(f"Title: {{article['title']}} ({{article['cl100k_tokens']}} tokens)")
     print(f"URL:   {{article['url']}}")
     print(f"Date:  {{article['date_published']}}")
+    print(f"Hero Image: {{article['featured_image_path']}}")
+    print(f"Screenshot: {{article['screenshot_path']}}")
     break
 ```