EPS Research RAG Corpus Series — FAISS Semantic Search Indexes v1.0

Pre-built FAISS semantic search indexes for all five corpora in the EPS Research RAG Corpus Series. Each index enables natural-language similarity search over structured astrophysical data without requiring re-embedding.

Embedding model: sentence-transformers/all-MiniLM-L6-v2 (384-dimensional, L2 distance)
Index type: faiss.IndexFlatL2 (exact search)
Total vectors: 2,064 across 5 corpora

Contents (15 files, 3 per corpus):

Each corpus has three files: .faiss (binary index), _ids.json (ordered ID list mapping index row to object ID), _texts.json (curated summary strings that were embedded, for inspection and debugging).

Quick start:

import faiss, json
from sentence_transformers import SentenceTransformer

index = faiss.read_index('v7_sparc.faiss')
ids = json.load(open('v7_sparc_ids.json'))
model = SentenceTransformer('all-MiniLM-L6-v2')

q = model.encode(['dwarf irregular low surface brightness']).astype('float32')
D, I = index.search(q, 5)
print([ids[i] for i in I[0]])

Full platform: https://github.com/eps-research/rag-corpus-series