AWS Documentation Chunk Embeddings: 89,221 Vectors for Semantic Search

89,221 document chunk embeddings derived from AWS public documentation. The dataset was created by author 'mouadja' and was last updated on 2026-05-28. Each row includes a text chunk, its file path, and a 512-dimensional vector generated using the OpenAI text-embedding-3-small model.

Use Cases

• Build a semantic search engine for AWS documentation based on the precomputed text embeddings.
• Fine-tune or evaluate embedding models for technical domain adaptation using the provided content and vector pairs.
• Create a retrieval-augmented generation (RAG) pipeline for AWS support chatbots based on the chunked documentation text.
• Analyze documentation structure and information density across different AWS services based on the chunk index and file path metadata.

Strengths

• Contains 89,221 precomputed document chunk embeddings, providing a substantial foundation for search applications.
• Uses the specific OpenAI text-embedding-3-small model with 512 dimensions, ensuring consistency for cosine similarity searches.
• Includes original text content and file path metadata, allowing traceability back to the source AWS documentation.

Limitations

• Row count for the underlying text data is unknown, which may limit suitability assessment for specific scale requirements.
• Column-level documentation beyond the provided schema is absent; field semantics for some columns like '_pinecone_id' must be inferred after download.
• Last updated 2026-05-28 15:09:26; freshness relative to the latest AWS documentation should be verified.

Provenance

Source

AWS public documentation.

Collection Method

Documentation was chunked and embedded using the OpenAI text-embedding-3-small model.

Freshness

Last updated 2026-05-28 15:09:26.