skills.rest
eddielueng
eddieluengCOMMUNITY
Explore all skills by this author →
GitHub

faiss

Build FAISS indexes and perform k-nearest-neighbor retrieval on dense vector embeddings.

1|Updated Apr 14, 2026
Search Tags:#rag#vector search#similarity search#faiss#gpu acceleration#nearest neighbors
One-click install
npx skills add https://github.com/eddielueng/hermes-agent-zh --skill faiss-eddielueng
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: faiss
Source: https://github.com/eddielueng/hermes-agent-zh/tree/main/optional-skills/mlops/faiss
Command: npx skills add https://github.com/eddielueng/hermes-agent-zh --skill faiss-eddielueng

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires faiss-cpu, faiss-gpu, numpy, and includes references (resource) components.

What problem does it solve?

FAISS solves the problem of performing fast similarity search over large collections of dense vector embeddings when you need low-latency nearest-neighbor retrieval.

Core Features & Use Cases

  • High-performance vector indexing: Build and query multiple FAISS index types for exact or approximate search.
  • Scales to massive datasets: Support very large vector corpora with GPU acceleration and billion-scale retrieval patterns.
  • Practical retrieval pipelines: Use FAISS directly or via common ecosystem integrations such as LangChain and LlamaIndex.

Use cases: semantic search over embeddings, recommendation candidate retrieval, offline/batch embedding retrieval, and high-throughput RAG retrieval where metadata filtering is not the primary requirement.

Quick Start

Use the faiss skill to install FAISS for CPU or GPU, create an in-memory vector index, add your embeddings, and run a k-nearest-neighbors similarity search.

Frequently Asked Questions about faiss

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I perform fast similarity search over large vector embeddings?▼

Fast similarity search over large vector embeddings is achieved by building FAISS indexes for exact or approximate nearest-neighbor retrieval. You add dense vectors to an in-memory index and run k-nearest-neighbors queries for low-latency results.

What is the best way to scale vector retrieval for billion-scale datasets?▼

Scaling vector retrieval to billion-scale datasets requires approximate index types like IVF or PQ combined with GPU acceleration. FAISS enables high-throughput retrieval by building specialized indexes and utilizing faiss-gpu for massive parallel processing.

Can I use FAISS for semantic search and RAG pipelines?▼

Yes, FAISS works for semantic search and RAG pipelines by performing k-nearest-neighbor retrieval on dense vector embeddings. It integrates with frameworks like LangChain and LlamaIndex to provide high-throughput candidate retrieval for embedding-driven workflows.

Do I need to normalize vectors for cosine similarity in FAISS?▼

Yes, vector normalization is required when using inner-product indexes for cosine similarity in FAISS. Normalizing dense embeddings before adding them to the index ensures accurate similarity measurements during the nearest-neighbor retrieval process.

What are the limitations of using FAISS for vector search?▼

A limitation of FAISS for vector search is that metadata filtering is not its primary focus. While it excels at high-throughput dense vector retrieval, complex attribute filtering alongside approximate nearest-neighbor search requires additional pipeline components.

Related Skills

Explore More Data & Analytics Skills→
afel6afel6

faiss

Build and query FAISS indexes for vector similarity search over dense embeddings.

Community
Advanced
ldzhhxxldzhhxx
1

qdrant-vector-search

Index vectors from data/vectors.csv and search for similar vectors.

Community
Advanced
cxnaivecxnaive

chroma

Store and retrieve vector embeddings with metadata filtering for RAG applications.

Community
Intermediate
SignmanalSignmanal
1

qdrant-vector-search

Implement Rust-powered HNSW vector similarity search with hybrid filtering.

Community
Advanced