Awesome GitHub RepositoriesMultimodal Document Processing

LightRAG is a graph-based retrieval framework designed to build retrieval-augmented generation pipelines. It structures unstructured text into knowledge graphs, enabling multi-hop reasoning and complex query synthesis across large document collections. By integrating dense vector embeddings with structured knowledge graphs, the system facilitates both similarity-based and relationship-aware information retrieval. The framework distinguishes itself through a dual-level retrieval strategy that combines low-level keyword matching with high-level semantic graph traversal to capture both specific

Sglang is a high-performance inference engine and serving system designed for large language and multimodal models. It provides a programmable interface for orchestrating complex generation workflows, enabling developers to coordinate multi-turn dialogues, tool invocations, and reasoning chains through a domain-specific language. The platform is built to support production-scale deployments, offering an OpenAI-compatible API that allows for integration with existing application ecosystems. The system distinguishes itself through a disaggregated architecture that separates compute-intensive pr

Vercel is a cloud platform for building, deploying, and scaling web applications. It provides a unified infrastructure that automates the build process by detecting project frameworks and distributing static and dynamic content through a global content delivery network. The platform executes application logic using serverless functions that scale automatically based on real-time traffic demand. The platform distinguishes itself through a centralized AI gateway that proxies requests to multiple model providers, enabling standardized authentication, observability, and cost tracking. It supports

Instructor is a framework designed for structured data extraction, validation, and language model integration. It functions as a library that transforms unstructured text into validated, type-safe objects by leveraging schema definitions and model-specific tool-calling capabilities. By acting as a validation middleware, the project ensures that language model outputs strictly conform to defined data structures. The library distinguishes itself through a robust validation-based retry loop that automatically re-submits failed responses with error feedback to iteratively correct schema complianc

Paper-qa is a retrieval augmented generation system designed for question answering and analysis of scientific literature and technical documents. It functions as an LLM-powered research assistant that extracts grounded answers and summaries with citations from a document library. The system utilizes an agentic RAG orchestrator to iteratively refine search queries and gather evidence through multi-step tool calling. It features a multimodal document parser that extracts text, tables, and images from PDFs, alongside a vector-based indexer that embeds and caches document libraries for efficient

This project is a reference implementation and application template for Retrieval-Augmented Generation (RAG). It integrates Azure OpenAI with Azure AI Search to enable conversational chat interfaces that provide grounded responses based on private enterprise data. The system is distinguished by its multimodal AI interface, allowing it to process and reason over combined text, image, and PDF content. It employs a hybrid search architecture that combines vector and keyword retrieval with semantic reranking to prioritize the most relevant documents for prompt augmentation. The project covers a

BERTopic is a topic modeling library used to extract interpretable themes from collections of text documents and images. It functions as a document clustering framework that transforms unstructured data into numerical vectors to group semantically similar content. The project distinguishes itself through a multimodal embedding tool that allows for joint clustering of text and images in a shared vector space. It also features a class-based TF-IDF representation engine to identify representative words for clusters and an integrated system for using large language models to generate natural lang

MNBVC is a dataset pipeline and toolkit designed for the collection, cleaning, and normalization of massive text and code corpora used to train large language models. It provides specialized tools for harvesting source code, commit histories, and repository metadata from version control platforms, alongside a multilingual text corpus collector for gathering parallel text and academic papers. The project distinguishes itself through comprehensive capabilities for processing diverse document types, including a PDF-to-text converter that transforms complex layouts and formulas into structured JS

ruby_llm is an LLM integration framework and AI agent orchestrator designed to connect applications to multiple large language model providers through a unified interface. It serves as a toolkit for building autonomous assistants with custom personas, managing structured output via JSON schemas, and implementing vector embedding engines for semantic search. The project distinguishes itself as an observability suite and multimodal toolkit. It provides specialized capabilities for tracking token usage, calculating model costs, and tracing workflows via OpenTelemetry, while supporting the proces

El kit de datos sintéticos es un framework integrado diseñado para generar, curar y formatear conjuntos de datos de entrenamiento para modelos de lenguaje. Proporciona un pipeline integral que transforma documentos fuente sin procesar en datos estructurados adecuados para el ajuste fino, el razonamiento y el entrenamiento de modelos de uso de herramientas. El framework se distingue por un motor de orquestación modular que gestiona todo el ciclo de vida de la preparación de datos. Admite entrada multimodal extrayendo contenido de texto e imagen de varios formatos de archivo, mientras emplea fragmentación consciente del contexto para mantener la coherencia semántica. El proceso de generación es impulsado por la inyección de prompts basada en plantillas, y la salida resultante se valida a través de un sistema de evaluación automatizado que utiliza modelos de lenguaje como jueces para asegurar la calidad y la precisión. El proyecto cubre una amplia gama de capacidades de procesamiento de datos, incluyendo análisis de documentos, filtrado de calidad automatizado y serialización agnóstica al esquema. Admite la creación de diversos ejemplos de entrenamiento, como trazas de razonamiento y demostraciones de uso de herramientas, y exporta los conjuntos de datos finales a formatos estandarizados para su compatibilidad con frameworks de entrenamiento de machine learning. Los usuarios gestionan el flujo de trabajo de generación y las etapas del pipeline a través de archivos de configuración centralizados y argumentos de línea de comandos.