infiniflow/infinity

Features

• Hybrid Vector and Full-Text Search - Combines dense vectors, sparse vectors, and full-text indices into a single result set using reranking expressions.
• Multimodal Data Storage - Functions as a multimodal vector store capable of maintaining strings, numerics, and vectors in a single system.
• Distributed Architectures - A scalable distributed vector database architecture supporting cluster deployment for high availability and large datasets.
• Low-Latency Vector Retrieval - Provides a distributed architecture optimized for low-latency retrieval of large-scale vector datasets.
• Data Querying and Selection - Provides capabilities for querying structured data using filters, sorting, grouping, and aggregation.
• Database Record Management - Creates tables and performs basic data operations to insert, update, delete, import, and export records.
• Distributed Database Clusters - Organizes nodes into a coordinated group to ensure high availability and scalability for large datasets.
• Full Text Search - Ships a full-text search engine to identify relevant rows based on text matches and keyword highlighting.
• Hybrid Search Engines - Combines semantic vector retrieval with traditional keyword-based indexing and reranking algorithms.
• Multimodal Stores - A multimodal storage system for managing diverse data types including tensors, strings, and numerics.
• Data Manipulation Operations - Implements comprehensive data manipulation operations including inserting, updating, deleting, importing, and exporting rows.
• Row Insertions - Inserts, updates, and deletes rows of data within specified tables to maintain current records.
• Database Indexes - Implements database indexes on columns to accelerate retrieval and reduce query latency.
• Programmatic Table and Index Managements - Provides programmatic management for creating, altering, and deleting table indexes.
• Vector Databases - An optimized database for storing and querying the dense and sparse vectors used by LLM applications.
• Vector Search - Performs similarity search on dense and sparse vectors using mathematical distance metrics.
• Dense Vector Retrieval - Identifies the closest rows to dense vectors or expands tensors for embedding-based retrieval.
• Distributed Database Deployment - Supports the deployment of distributed database clusters with coordinator and worker nodes for high availability.
• Vector Distance Metrics - Calculates mathematical distance between query vectors and stored embeddings to identify the closest matching records.
• RAG Pipelines - Provides the indexing and reranking infrastructure necessary to build retrieval augmented generation workflows.
• Vector Store Backends - Serves as a high-performance storage and retrieval backend for multimodal data in LLM applications.
• Result Reranking - Implements algorithms to re-order hybrid search results to improve the precision and relevance of the top hits.
• Multi-Vector Embeddings - Stores two dimensional tensors and expands them into fixed dimension embedding vectors during ingestion.
• Sparse Embeddings - Implements efficient retrieval of high-dimensional sparse vector representations for keyword-based similarity search.
• Point-in-Time Database Snapshots - Creates and restores point-in-time system or table backups to preserve and recover database state.
• Data Import and Export - Moves datasets into the system from external files or extracts stored data for other applications.
• Embedding Conversions - Transforms two dimensional tensor data into fixed dimension embedding vectors during the data insertion process.
• Data Storage Optimizers - Includes storage optimization tools to compact tables and optimize indexes for better efficiency.
• Database and Table Management - Enables the creation and organization of databases and tables for structured data storage.
• Database Backup and Restore - Implements systems for creating and restoring data backups to prevent data loss.
• Database Client SDKs - Performs data operations and system management through a Python library for streamlined integration.
• Database HTTP APIs - Executes database operations and manages data using standardized web endpoints for language agnostic integration.
• Database Schema Management - Provides utilities for defining and managing database schemas through column and index definitions.
• Column Indexing - Implements specialized column-based indexing to accelerate retrieval and reduce query latency.
• Database Management APIs - Provides HTTP APIs for managing database server configuration and record operations.
• Cluster Node Management - Includes tools for managing cluster node lifecycles, roles, and member status.
• Docker Deployments - Provides containerized deployment via Docker for consistent environment setup of server and client processes.
• Self-Hosted Database Deployments - Provides pre-compiled binaries for self-hosting the database server on user-owned infrastructure.
• Single-Binary Deployments - Packages the entire database engine into a standalone executable to eliminate external runtime dependencies.
• Database SDKs - Manipulates database records and interacts with the system using a native programming interface.
• Inference and Serving - AI-native database for hybrid search and retrieval.
• Inference Engines - AI-native database for fast hybrid search.
• Vector Databases - AI-native database for hybrid vector and full-text search.
• Database Systems - AI-native database for vector and full-text search.
• Bases de datos - AI-native database for LLM applications.
• Vector Databases - AI-native database for vector and full-text search.

Historial de estrellas