README.md

---
license: mit
task_categories:
  - text-retrieval
language:
  - en
size_categories:
  - 10K<n<100K
---

# CAG-Lab AWS Docs Vectors

89,221 document chunk embeddings from AWS public documentation.

- **Embedding model**: OpenAI `text-embedding-3-small` (512 dimensions)
- **Distance metric**: Cosine
- **Source**: AWS public documentation chunks

## Schema

| Column | Type | Description |
|--------|------|-------------|
| id | string | Deterministic UUID |
| embedding | list[float32] | 512-dim vector |
| content | string | Document chunk text |
| filePath | string | Original file path |
| chunkIndex | string | Chunk position |
| _pinecone_id | string | Original Pinecone vector ID |

## Usage

```python
from datasets import load_dataset

ds = load_dataset("mouadja/aws-docs")
```

Or use with CAG-Lab:

```bash
python scripts/setup_vectordb.py
```