The most accurate document search and store for building AI apps

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

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

Vision-Augmented Retrieval and Generation (VARAG) - Vision first RAG Engine

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

High-Performance Engine for Multi-Vector Search

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

🐊 Snappy's unique approach unifies vision-language late interaction with structured OCR for region-level knowledge retrieval. Like the project? Drop a star! ⭐

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

A monorepo containing various utility scripts, tools, and applications for development, automation, and AI-powered tasks.

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.

The PyQt6 application using ColPali and OpenAI to show Efficient Document Retrieval with Vision Language Models

The repo provides the code for Qdrant for efficient image indexing and retrieval using models such as ColPali, ColQwen, and VDR-2B-Multi-V1, jina embeddings v4 etc enhancing multimodal search capabilities across various applications.

REST API for computing cross-modal similarity between images and text using the ColPaLI vision-language model

Vision-based RAG

Engineering Homework solver using ColPali PDF retrieval, Qwen2.5-VL Multimodal analysis, and DeepSeek code generation with schemdraw to create step-by-step solutions with circuit diagrams.

A high-performance RAG system for PDFs using multi-vector embeddings (ColPali / ColQwen / ColSmol) with vector search in Qdrant, prefetch optimization, and reranking for improved relevance. Designed for speed, accuracy, and scalability, this system is ideal for building intelligent search, document understanding, and QA applications.

VisRAG playground for finding your perfect embedding + VLM combo. Index PDFs with multimodal models, compare responses side-by-side.

Fast, memory-efficient retrieval-augmented generation over visually rich documents. Powered by HPC-ColPali (Hierarchical Patch Compression for ColPali) and Janus-Pro.

RAG Lab — Local visual document analysis workbench powered by ColPali embeddings and Ollama vision models