Colpali Python Packages

colpali-engine

The code used to train and run inference with the ColVision models, e.g. ColPali, ColQwen2, and ColSmol.

fast-plaid

High-Performance Engine for Multi-Vector Search

optivision-rag

Visual-token compression for VLM document retrieval: 448 KB to 3.95 KB per page (113.5x) at 86.7% of baseline nDCG@5. ColPali/ColSmol, PyTorch, Qdrant.

byaldi

Use late-interaction multi-modal models such as ColPali in just a few lines of code.

late-interaction-kernels

Fused Triton kernels for late-interaction (MaxSim) scoring

vllm-factory

Production inference for encoder models - ColBERT, GLiNER, ColPali, embeddings etc. - as vLLM plugins for online and in-process deployment

vidore-benchmark

Vision Document Retrieval (ViDoRe): Benchmark. Evaluation code for the ColPali paper.

latence-solver

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.

voyager-index

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.

litepali

LitePali is a minimal, efficient implementation of ColPali for image retrieval and indexing, optimized for cloud deployment.

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.

ragpipe-lite

`ragpipe` is a lightweight RAG ingestion pipeline toolkit. It helps you go from documents → chunks → embeddings → vector store export.