elips 1.0.0

Embedded local vector database (SQLite for vectors)

ELIPS (Python Bindings)

Embedded Local Index & Persistence System (ELIPS) Python package.

ELIPS is an in-process vector and document retrieval engine built in C++23 with native Python bindings. It keeps the embedded deployment model of SQLite, but adds ANN indexes, typed metadata filters, first-class document lineage, hybrid retrieval, segmented persistence, and optional GPU-backed indexes.

Installation

From PyPI

pip install elips

From Source

pip install .

Note: Building from source requires a C++23 compatible compiler (GCC 13+ or Clang 17+), CMake >= 3.24, and Ninja.

Quick Start

import elips

# Connect to database (either path or ":memory:")
engine = elips.connect(":memory:", dimension=128)

# Access or create a documents arena
arena = engine.arena("documents")

# Ingest documents with text and optional metadata
arena.ingest(
    texts=["alpha design note", "beta incident runbook"],
    meta=[{"kind": "design"}, {"kind": "ops"}],
)

# Run semantic search
for hit in arena.probe_text("alpha", top=2):
    print(hit.key, hit.distance, hit.text, hit.meta)

Features

• Vector Search: Optimized HNSW (graph) and exact index backends.
• Embedded Deployment: Simple SQLite-like single-writer/multi-reader lock model.
• Local Embedding: Built-in local text embedding using local models.
• Metadata Filters: Rich metadata equality filters via MetadataIndex and Filter.
• Hybrid Retrieval: Combine dense semantic search and document retrieval out of the box.
• Segmented Persistence: Durable writes with WAL crash recovery and snapshot checkpoints.
• GPU Acceleration: Optional Metal backend on Apple Silicon and CUDA/Vulkan backends on Linux.

License

ELIPS Python bindings are released under the MIT License.