farhanlodi31 

High-accuracy native .NET OCR powered by EasyOCR's neural models running on ONNX Runtime. No Python required.

CUDA GPU execution provider for EasyOcrSharp. Adds Microsoft.ML.OnnxRuntime.Gpu so detection and recognition run on NVIDIA GPUs.

Fast, lightweight, and flexible object-to-object mapper for .NET with support for nested objects, lists, enums, and sub-object injection.

Universal, manifest-driven, pure-managed model inference for .NET. All-in-one: ONNX loader, multi-dtype CPU engine, tensors, pipeline, plus audio (FFT/log-mel) and text (WordPiece) front ends — zero external,... More information

High-performance persistent collection library with ACID transactions, indexing, and advanced data structures

ImageSharp-based image pre/post-processing adapter for ModelSharp — resize/normalize tensor preprocessing and classification/detection postprocessing.

Model acquisition for ModelSharp — download and cache ONNX / GGUF / safetensors models (and their external-data + tokenizer files) from Hugging Face Hub or any URL, then run them in one line via... More information

ILGPU GPU execution backend for ModelSharp — C# kernels JIT-compiled to CUDA / OpenCL, with a CPU fallback and whole-graph GPU dispatch.

A RAG (Retrieval-Augmented Generation) framework for .NET. Build intelligent applications with document ingestion, text chunking, embedding generation, vector storage, and semantic search. Includes all RAGify... More information

Production-ready hybrid post-quantum cryptography library for .NET authentication. Combines classical ECDSA P-256 with post-quantum ML-DSA-65 (FIPS 204) to provide quantum-resistant JWT signing and validation.... More information

Core interfaces and contracts for RAGify. Provides the foundation for building custom implementations of embedding providers, vector stores, chunking strategies, and retrieval engines. Use this package when... More information

Core domain models and shared utilities for RAGify. Contains common data structures, models, and helper classes used across all RAGify components including document models, chunk metadata, embedding vectors,... More information

Embedding provider implementations for RAGify. Convert text into high-dimensional vectors using popular embedding models including OpenAI, Azure OpenAI, Ollama, ONNX (local models), Hugging Face, Cohere,... More information

Vector storage implementations for RAGify. Store and search high-dimensional embeddings efficiently with support for in-memory storage, PostgreSQL (pgvector), Pinecone, and Weaviate. Includes extensible... More information

Retrieval engine for RAGify. Perform semantic search by embedding user queries and finding the most relevant document chunks using similarity search. Includes support for metadata filtering, result ranking, and... More information

Intelligent text chunking strategies for RAGify. Break down large documents into optimal-sized chunks for embedding and retrieval. Includes fixed-size chunking, sentence-aware chunking, and sliding window... More information

Document ingestion services for RAGify. Extract and process text from various document formats including PDF, Microsoft Word (DOCX), Excel (XLSX), HTML, and plain text files. Automatically handles file parsing... More information

CUDA GPU execution provider for PaddleOcrNet. Adds Microsoft.ML.OnnxRuntime.Gpu so detection, classification and recognition run on NVIDIA GPUs.

High-accuracy native .NET OCR powered by PaddleOCR's neural models (DB detection, text-line orientation, SVTR/CRNN recognition) running on ONNX Runtime. No Python required.