File size: 3,935 Bytes
f95a2b3 1de52b8 f95a2b3 1de52b8 f95a2b3 1de52b8 f95a2b3 1de52b8 f95a2b3 1de52b8 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 | # Epstein Vectorized Database
This repository contains a **vectorized derivative** of publicly released U.S. House Oversight Committee Epstein estate documents. The database provides efficient semantic search and retrieval-augmented generation (RAG) capabilities via FAISS embeddings and metadata.
## Database Contents
- `epstein_index.faiss`: FAISS vector index (384-dim, normalized embeddings via all-MiniLM-L6-v2, IndexFlatIP for cosine similarity)
- `epstein_metadata.parquet`: Metadata for each vector (id, filename, text_snippet, chars, words)
- `epstein_metadata.jsonl`: Same metadata in JSONL format
**Total documents indexed: 25,800**
## Source Attribution
**Original source:** U.S. House Committee on Oversight and Government Reform public release "Oversight Committee Releases Additional Epstein Estate Documents" (November 12, 2025):
[https://oversight.house.gov/release/oversight-committee-releases-additional-epstein-estate-documents/](https://oversight.house.gov/release/oversight-committee-releases-additional-epstein-estate-documents/)
**Original dataset on HuggingFace:**
[tensonaut/EPSTEIN_FILES_20K](https://huggingface.co/datasets/tensonaut/EPSTEIN_FILES_20K)
## Usage
### Load the index and metadata:
```python
import faiss
import pandas as pd
index = faiss.read_index('epstein_index.faiss')
meta = pd.read_parquet('epstein_metadata.parquet')
```
### Semantic search example:
```python
from sentence_transformers import SentenceTransformer
import numpy as np
model = SentenceTransformer('all-MiniLM-L6-v2')
query = "your search query"
q_emb = model.encode([query], convert_to_numpy=True)
q_emb = q_emb / np.linalg.norm(q_emb, axis=1, keepdims=True)
D, I = index.search(q_emb.astype('float32'), k=10) # top-10 results
results = meta.iloc[I[0]]
```
## Usage Responsibilities
Users are responsible for:
- Using the dataset only for **lawful purposes** and in accordance with institutional and ethical review requirements
- Treating individuals mentioned in the documents with respect and avoiding sensationalism or misuse of sensitive material
- **Clearly distinguishing model-generated content** from verified facts, and citing primary sources appropriately
- Complying with applicable copyright law, privacy law, and institutional policies
### NOT intended for:
- Fine-tuning language models without explicit legal review
- Harassment, doxing, or targeted attacks
- Attempts to deanonymize or circumvent existing redactions
- Making or amplifying unverified allegations as factual claims
## Content Warning
The underlying corpus contains sensitive material related to:
- Sexual abuse and exploitation
- Trafficking and violence
- Unverified allegations, opinions, and speculation
Readers should approach with care and appropriate context.
## Legal Disclaimer (Non-Authoritative)
- This is a **derivative vectorized index only**; the original documents are copyrighted by their respective authors and the U.S. House Committee on Oversight and Government Reform
- This dataset does **not grant any license** to reproduce or distribute the underlying documents beyond what may be permitted by law (fair use, etc.)
- Users are **solely responsible** for ensuring compliance with copyright, privacy, and institutional policies
- **Seek independent legal counsel** if using this corpus in a public-facing product or for model training at scale
## Files
- `epstein_index.faiss`: FAISS IndexFlatIP (384-dim, normalized vectors)
- `epstein_metadata.parquet`: Parquet metadata table
- `epstein_metadata.jsonl`: JSONL metadata (one record per line)
## License
This vectorized database is provided under the same legal constraints as the original House release. See "Usage Responsibilities" and "Legal Disclaimer" above. No original content is reproduced; only embeddings and metadata snippets are stored.
|