EPRI Public Reports RAG Database: Chunked Texts with Embeddings

EPRI Public Reports RAG Database is an open-source SQLite database containing EPRI public publications, shared as chunked texts with dense embeddings created using BGE-M3. The artifact is distributed via Hugging Face and EPRI.com to enable external organizations to ingest these documents into internal RAG pipelines. The dataset was last updated on 2026-06-16.

Use Cases

• Implementing semantic search over EPRI technical publications based on the provided dense embeddings.
• Augmenting large language model responses with factual information from energy research reports.
• Building internal knowledge bases for energy utilities or consultancies using pre-processed document chunks.

Strengths

• Includes dense embeddings generated with the BGE-M3 model, facilitating immediate use in vector search.
• Artifact is distributed as a downloadable SQLite database, simplifying integration.

Limitations

• Description metadata is limited; actual data quality requires manual inspection after download.
• Row count and column-level documentation are absent; field semantics must be inferred after download.

Provenance

Source

EPRI (Electric Power Research Institute)

Collection Method

Likely involves processing and chunking of EPRI's publicly available publications.

Freshness

Last updated 2026-06-16 15:16:16; freshness should be verified.