Explore the riches of an edge native RAG pipeline!!!

[biapalmeiro]

Using cloud hosted RAG pipelines to query and understand large corpuses of sensitive data like enterprise documents is not a feasible approach. Deploying RAG solutions on enterprise internal compute cluster is challenging due to the complexity involved – creating the RAG pipeline, ingestion of different document types, selecting right models, using the right model serving techniques etc.

Unlock the potential of your private data corpus with the Chat Question and Answer application—an edge-native RAG solution designed for seamless adaptation, customization, and deployment

The sample application provides the following distinct capabilities to its users:

• Uses popular GenAI framework like LangChain to chain different parts of the RAG pipeline in an industry standard approach
• Provides a rich catalogue of customizable microservices to help with data ingestion and model inferencing.
• Pre-validated on recommended Intel® AI capable hardware portfolio including the latest dGPU portfolio for a seamless day 1 experience
• Several customization knobs like LLMs to be used, document formats supported, model servings used etc.
• Edge native application that can be deployed in resource constrained environments through the “core” version of the sample application
• Uses optimized Intel® resources like OpenVINO, OVMS, etc. to guarantee optimal performance on Intel® HW.

Ready to play with the RAG pipeline on your private and sensitive data corpus? Checkout the sample application versions here: Modular (https://lnkd.in/dWcWcB9M) and Core (https://lnkd.in/ddsaK3Rt).

#GenAI #RAG #Chat #optimizedRAGpipeline #OpenEdgePlatform #EdgeAILibraries #IamIntel

Explore more edge-optimized tools in our GitHub ecosystem — designed to simplify deployment and speed up time-to-value in real-world applications.

Original Bartosz Ciepluch’s Post