# Hugging Face Dataset Setup and Deployment Guide

This document describes how to publish the Anvil Index RAG Packs sample to Hugging Face Datasets Hub.

## Prerequisites

- Hugging Face account (create at https://huggingface.co)
- Hugging Face CLI installed (`pip install huggingface-hub`)
- Write access to anvilindex organization on Hugging Face
- Git LFS installed (for large files, if applicable)

## Directory Structure

```
huggingface-sample/
├── README.md                 # Dataset card with YAML frontmatter
├── CHUNKS.md                 # Quick reference index
├── SETUP.md                  # This file
└── sample_chunks/
    ├── governance/           # 8 files (4 chunks)
    ├── security/            # 8 files (4 chunks)
    ├── privacy/             # 8 files (4 chunks)
    ├── finserv/             # 8 files (4 chunks)
    └── healthcare/          # 8 files (4 chunks)
```

Total: 42 files (40 chunk files + 2 documentation files)

## Step 1: Prepare Hugging Face Repository

### Option A: Using Hugging Face Web Interface

1. Go to https://huggingface.co/datasets
2. Click "Create new dataset"
3. Fill in:
   - Owner: anvilindex (or your organization)
   - Dataset name: anvil-index-rag-packs-sample
   - Description: "20-chunk sample showcasing Anvil Index RAG Packs quality and schema"
   - License: Creative Commons Attribution 4.0 International
   - Make dataset public
4. Click "Create dataset"

### Option B: Using Hugging Face CLI

```bash
huggingface-cli login
huggingface-cli repo create \
  --repo-type dataset \
  --repo-name anvil-index-rag-packs-sample \
  --organization anvilindex \
  --public
```

## Step 2: Clone and Configure Repository

```bash
# Clone the repository
cd /tmp
git clone https://huggingface.co/datasets/anvilindex/anvil-index-rag-packs-sample
cd anvil-index-rag-packs-sample

# Configure Git
git config user.name "Anvil Index"
git config user.email "support@anvilindex.com"
```

## Step 3: Add Dataset Files

```bash
# Copy all files from local directory
cp -r /sessions/pensive-epic-babbage/mnt/Domain-Specific\ Agentic\ Knowledge\ Bases\ \(RAG\ Packs\)/huggingface-sample/* .

# Verify structure
ls -la
tree sample_chunks/
```

## Step 4: Create .gitattributes for Large Files (if needed)

If any .json files exceed 100MB (they won't in this sample), configure Git LFS:

```bash
# Create .gitattributes
echo "*.json filter=lfs diff=lfs merge=lfs -text" >> .gitattributes
echo "*.md filter=lfs diff=lfs merge=lfs -text" >> .gitattributes

# Add to git
git add .gitattributes
git commit -m "Configure Git LFS for large files"
```

## Step 5: Commit and Push to Hugging Face

```bash
# Stage all files
git add -A

# Commit
git commit -m "Add 20-chunk Anvil Index RAG Packs sample dataset

This sample includes 4 representative chunks from each of 5 regulatory domains:
- AI Governance (4 chunks)
- AI Security (4 chunks)
- AI Privacy (4 chunks)
- Financial Services (4 chunks)
- Healthcare (4 chunks)

Each chunk includes both markdown content and rich JSON metadata with:
- Semantic titles and controlled vocabularies
- Cross-references to related chunks across domains
- Compliance metadata (obligations, enforcement dates, risk tiers)
- AI lifecycle and actor role classifications

This is a free lead magnet showcasing Anvil Index quality and schema.
License: CC BY 4.0"

# Push to Hugging Face
git push origin main
```

## Step 6: Verify on Hugging Face

1. Navigate to https://huggingface.co/datasets/anvilindex/anvil-index-rag-packs-sample
2. Verify:
   - README.md displays correctly with dataset card information
   - CHUNKS.md appears in the file browser
   - sample_chunks/ folder structure is visible
   - All 40 chunk files are present

## Step 7: Update Dataset Card Settings

Go to the dataset repo settings (https://huggingface.co/datasets/anvilindex/anvil-index-rag-packs-sample/settings):

### Basic Information
- **Dataset name**: anvil-index-rag-packs-sample
- **Description**: Free 20-chunk sample of Anvil Index RAG Packs showcasing regulatory compliance data quality and schema
- **License**: CC-BY-4.0
- **Tags**: rag, regulatory-compliance, ai-governance, ai-security, legal-tech, vector-database

### Visibility
- Make dataset public (already set via web interface)

### Citation
Add this to the README YAML if not already included:

```yaml
citation_info:
  url: "https://huggingface.co/datasets/anvilindex/anvil-index-rag-packs-sample"
```

## Step 8: Create Dataset Card on Hugging Face (Alternative)

If you prefer to edit directly on Hugging Face:

1. Go to dataset page
2. Click "Edit dataset card"
3. Paste the contents of README.md
4. Ensure YAML frontmatter is at the top
5. Save

## Post-Launch: Monitoring and Updates

### Track Downloads
- Hugging Face provides download statistics in the dataset repo
- Monitor for spikes indicating lead magnet effectiveness

### Handle Issues
- Set up discussions for questions about the sample
- Monitor for feature requests (e.g., "Can you add X regulatory domain?")
- Track conversion from sample to paid product

### Future Updates
- Version 1.1: Add CHUNKS.md searchable index
- Version 1.2: Add optional Parquet exports for vector DB bulk loading
- Version 2.0: Refresh to include updates from Pack version 1.3.0

## Integration with Product Sales

### Lead Magnet Workflow
1. User downloads sample from Hugging Face
2. User evaluates in their vector DB
3. User visits https://anvilindex.com/products
4. User selects and purchases full pack

### Marketing Copy Suggestions
- "Download free 20-chunk Anvil Index sample on Hugging Face"
- "Evaluate RAG Pack quality before purchasing"
- "See how regulatory cross-references work"
- "Test in your vector database at no cost"

### Cross-Promotion
- Add Hugging Face dataset link to website footer
- Include in product announcement emails
- Share in AI/ML communities (Reddit, Twitter, LinkedIn)
- Submit to regulatory tech and legal tech newsletters

## Troubleshooting

### Large File Push Fails
Install Git LFS:
```bash
git lfs install
git lfs track "*.json" "*.md"
git add .gitattributes
git commit -m "Add LFS tracking"
git push origin main
```

### Permission Denied on Push
Ensure you are authenticated:
```bash
huggingface-cli login
git push origin main
```

### YAML Frontmatter Not Rendering
- Verify README.md begins with `---`
- Ensure YAML is valid (correct spacing, no tabs)
- Test at https://www.yamllint.com/

## References

- Hugging Face Datasets Documentation: https://huggingface.co/docs/datasets
- Dataset Card Guide: https://github.com/huggingface/datasets/blob/main/src/datasets/commands/upload_dataset.py
- Markdown Guide: https://huggingface.co/docs/hub/datasets-markdown

## Questions?

Contact: support@anvilindex.com

---

**Setup guide version**: 1.0.0
**Last updated**: March 29, 2026
**Tested with**: Hugging Face Hub v0.19+, Git 2.30+