late-interaction 开源项目推荐（34 个）

superlinear-ai /raglite

SYS RAGLite是一个用于使用DuckDB或PostgreSQL的检索增强生成（RAG）的Python工具包

topk-io /topk

High-quality search for AI-native applications.

lightonai /pylate-rs

PyLate efficient inference engine

tusKANNy /tachiom

Official repository of TACHIOM.

ejaasaari /lemur

[ICML'26] LEMUR reduces multi-vector retrieval for late interaction models such as ColBERT into regular single-vector retrieval.

hcompai /late-interaction-kernels

Fused Triton kernels for late-interaction (MaxSim) scoring

ddickmann /colsearch

High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.

enricollen /it-colbert

Monolingual Italian late-interaction (ColBERT) retriever for search and RAG

Ryenhails /NanoVDR

Efficient visual document retrieval: NanoVDR, DistilVDR and ColNanoVDR. Small query and document encoders distilled from frozen VLM retrievers.

maxbaluev /accreted-intelligence

AccInt - local-first MCP Work Model for coding agents that learns from real outcomes.

joe32140 /nano-plaid

Late-interaction retrieval (ColBERT + PLAID) in one numpy file — nanoGPT-style, with a Rust SIMD kernel ladder

pau-mensa /lateweave

Composable candidate generation and optional late-interaction (MaxSim) reranking, in Rust with Python bindings

mrsladoje /apollo

A physics-grounded, agent-driven digital twin for HP Metal Jet S100 3D printer

smarthi /pymuvera

Python library for MUVERA multi-vector retrieval via Fixed Dimensional Encodings. ColBERT / ColQwen2 / ColQwen3.5 compatible.

anirudhlakhotia /subbit-multivector-retrieval

Sign-coded late-interaction retrieval at 8 B/token: code, paper and full reproduction recipe for ReNeuIR @ SIGIR 2026

thlurte /limbed

ColBERT inference in C++ (ONNX Runtime + tokenizers-cpp) without PyTorch.

cmaftuleac /memtrails

Retrieval and exploration for any text — e-mail, chat, notes, documents, news, commits, conversations. Searchable at once, on your own machine, in any language.

maxbaluev /colpali-retrieve

ColPali-style multimodal late-interaction retrieval (text → document-image patches via MaxSim) — numpy reference implementation

PatelDivyam23 /multimodal-rag-system

Multimodal RAG that retrieves document pages as images using ColQwen2 late-interaction search, fused with dense and lexical retrieval. Modality-routed generation with page-level citations. Runs on a 4GB laptop GPU.

AbdelrahmanhAli /rag-colbert-late-interaction

ColBERT late-interaction retrieval with RAGatouille: index passages and run token-level search with optional grounded answering.