================================================================================ ANVIL INDEX RAG PACKS HUGGING FACE SAMPLE Quick Navigation Index ================================================================================ PROJECT: Free 20-chunk dataset for Hugging Face showcasing Anvil Index quality CREATED: March 29, 2026 VERSION: 1.0.0 STATUS: Complete and ready for publication ================================================================================ DOCUMENTATION FILES (Read These First) ================================================================================ 1. README.md (START HERE) - Main dataset card for Hugging Face Hub - Complete schema documentation - Use cases and feature highlights - Purchase links and pricing - Licensing information Action: Read to understand dataset purpose and structure 2. CHUNKS.md - Quick reference table of all 20 chunks - Organized by domain - Shows source and type for each chunk - Getting started guide Action: Use to quickly locate a specific chunk 3. SETUP.md (TECHNICAL) - Step-by-step Hugging Face publication guide - Repository setup instructions - Git push procedures - Post-launch monitoring Action: Follow to publish dataset to Hugging Face 4. MANIFEST.txt (REFERENCE) - Complete creation record and verification log - Quality assurance checklist - Statistical summary - File inventory Action: Reference for technical details and QA 5. INDEX.txt (THIS FILE) - Navigation guide - Quick reference - File locations Action: Use to orient yourself in the dataset ================================================================================ SAMPLE CHUNKS (20 Total) ================================================================================ Location: sample_chunks/{domain}/ GOVERNANCE (4 chunks) eu_ai_act_article_006 - EU high-risk AI classification rules nist_rmf_1_1 - NIST risk management framework intro nist_600_1_govern_1_1 - NIST governance functions oecd_ai_dd_step_1_1 - OECD due diligence process SECURITY (4 chunks) owasp_llm01_definition - OWASP Prompt Injection vulnerability ir8596_govern - NIST secure AI development governance atlas_aml_m0000 - MITRE ATL&CK reconnaissance tactic cisa_best_practices_001 - CISA AI data security PRIVACY (4 chunks) EU-AIA-PRIV-ART10-001 - EU AI Act privacy requirements CA-SB53-DEF-001 - California SB-53 AI definitions EDPB-JO-2026-BG-001 - EDPB algorithmic discrimination CISA-SBOM-2025-001 - CISA software bill of materials FINANCIAL SERVICES (4 chunks) SR-11-7-001 - Federal Reserve model risk management OCC-MRM-001 - OCC banking AI oversight SEC-EXAM-2026-001 - SEC investment adviser procedures TREAS-AI-FS-001 - Treasury financial services AI HEALTHCARE (4 chunks) FDA-SAMD-001 - FDA AI/ML software as medical device FDA-MLT-001 - FDA machine learning transparency HHS-AI-STR-001 - HHS healthcare AI strategy EMA-AI-RP-001 - EMA healthcare AI reflection Each chunk includes: - {chunk_id}.md (content) - {chunk_id}_metadata.json (structured metadata) ================================================================================ KEY NUMBERS ================================================================================ Chunks: 20 Files: 44 total (20 .md + 20 .json + 4 doc files) Domains: 5 (governance, security, privacy, finserv, healthcare) Chunks per domain: 4 (consistent) Total size: 216K Metadata fields per chunk: 25-35 (average 30) Cross-references per chunk: 3-6 (average 4.5) Source documents: 13 Jurisdictions: 6 (EU, US Federal, California, OECD, NIST, OWASP) ================================================================================ USE CASES ================================================================================ 1. EVALUATE BEFORE PURCHASING - Download sample to assess quality - Load into your vector database - Test search and retrieval - Decide if full packs worth purchasing 2. PROTOTYPING - Test RAG pipeline with real regulatory chunks - Develop metadata filtering logic - Build compliance workflow - Evaluate cross-domain queries 3. RESEARCH - Understand RAG chunk design - Study regulatory content structure - Analyze cross-reference patterns - Examine controlled vocabularies 4. INTEGRATION TESTING - Test embedding models - Validate semantic search - Build filtering pipelines - Develop multi-domain queries 5. PROOF OF CONCEPT - Demonstrate RAG capabilities to stakeholders - Show compliance automation potential - Illustrate cross-regulatory intelligence - Build internal buy-in ================================================================================ QUICK START (3 STEPS) ================================================================================ 1. EXPLORE SAMPLE - Read README.md for overview - Pick a chunk from CHUNKS.md - Open sample_chunks/{domain}/{chunk_id}.md to see content - Open sample_chunks/{domain}/{chunk_id}_metadata.json to see metadata 2. TEST IN VECTOR DB - Load _metadata.json files into your vector database - Test embeddings with your model - Try metadata filtering queries - Evaluate search performance 3. PURCHASE FULL PACKS (When Ready) - Visit https://anvilindex.com/products - Choose packs by domain - Pricing: $49/pack, $149/3-pack, $229/all-5 - Download and integrate full datasets ================================================================================ METADATA SCHEMA SUMMARY ================================================================================ Core Fields (all chunks): chunk_id - Unique identifier semantic_title - Human-readable title domain - AI_Governance, AI_Security, etc. source_document - Official source name source_authority - Authoring organization jurisdiction - Geographic scope source_url - Link to original Compliance Fields: risk_tier - general, medium_risk, high_risk compliance_obligation - true/false enforcement_dates - Timeline of requirements ai_actors - Roles affected regulatory_domain - Functional area Linking Fields: cross_references - Links to 3-6 related chunks keywords - Searchable terms See README.md for complete schema documentation. ================================================================================ PUBLICATION PATHWAY ================================================================================ BEFORE PUBLISHING: 1. Review all files in this directory 2. Verify README.md is complete 3. Check SETUP.md instructions 4. Confirm all 44 files are present PUBLISH TO HUGGING FACE: 1. Create dataset at https://huggingface.co/datasets 2. Follow SETUP.md steps to push files 3. Verify files appear on platform 4. Test download functionality AFTER PUBLISHING: 1. Share link on website and social media 2. Monitor download statistics 3. Enable discussions for Q&A 4. Track conversion to paid products URL after publication: https://huggingface.co/datasets/anvilindex/anvil-index-rag-packs-sample ================================================================================ SUPPORT & CONTACT ================================================================================ Product questions: support@anvilindex.com Sales inquiries: sales@anvilindex.com Website: https://anvilindex.com Products: https://anvilindex.com/products After publication: Hugging Face discussions: [link will be available on dataset page] ================================================================================ LICENSE ================================================================================ Dataset license: CC BY 4.0 (Creative Commons Attribution) - Free to use, share, adapt - Requires attribution to Anvil Index - Can be used commercially Source document licenses: - OWASP: CC BY-SA 4.0 - Government documents: Public domain / fair use - All sources attributed in metadata ================================================================================ DOCUMENT REFERENCE ================================================================================ File organization: INDEX.txt (this file) README.md (dataset card) CHUNKS.md (chunk reference) SETUP.md (publication guide) MANIFEST.txt (detailed record) sample_chunks/ governance/ (4 chunks) security/ (4 chunks) privacy/ (4 chunks) finserv/ (4 chunks) healthcare/ (4 chunks) Total: 44 files, 216K ================================================================================ NEXT STEP ================================================================================ Start with README.md, then choose your next action: - To understand the dataset: Read README.md - To find a specific chunk: Check CHUNKS.md - To publish to Hugging Face: Follow SETUP.md - For technical details: Review MANIFEST.txt Questions? Contact support@anvilindex.com ================================================================================ Version: 1.0.0 Created: March 29, 2026 Status: Complete and ready for publication ================================================================================