rag and RAG-Driven-Generative-AI
These are ecosystem siblings—both are reference implementations of RAG pipelines that demonstrate how to integrate different vector databases (Deep Lake, Pinecone) and LLM frameworks (LlamaIndex, OpenAI, Hugging Face) into production-ready systems, serving as educational blueprints rather than competing tools.
About rag
NVIDIA-AI-Blueprints/rag
This NVIDIA RAG blueprint serves as a reference solution for a foundational Retrieval Augmented Generation (RAG) pipeline.
Combines NVIDIA NIM microservices with GPU-accelerated vector search (cuVS), hybrid retrieval with reranking, and multimodal content extraction supporting text, tables, charts, and images. Built on LangChain orchestration with pluggable vector databases (Milvus, Elasticsearch), optional vision language models, and guardrailing for enterprise compliance. Supports multiple deployment modes including Docker, Kubernetes with NIM Operator, and Python library integration with OpenAI-compatible APIs.
About RAG-Driven-Generative-AI
Denis2054/RAG-Driven-Generative-AI
This repository provides programs to build Retrieval Augmented Generation (RAG) code for Generative AI with LlamaIndex, Deep Lake, and Pinecone leveraging the power of OpenAI and Hugging Face models for generation and evaluation.
Covers practical implementation of vector embedding pipelines with multimodal data support (text and images), hallucination mitigation through traceable source grounding, and optimization techniques like adaptive RAG and human-in-the-loop refinement. Demonstrates scaling patterns for production RAG systems across multiple vector databases (Pinecone, Chroma, Deep Lake) while balancing fine-tuning trade-offs and implementing knowledge graph visualization for complex data structures.
Scores updated daily from GitHub, PyPI, and npm data. How scores work